In this work the authors try to bring together recent regulation efforts by the European Union and first proposals for AI guidelines with recent trends in research: data and model cards. 

They propose the use of standardized cards to document AI applications throughout the development process. 

They introduce and argue for use-case and operation cards, along with updates for data and model cards to cope with regulatory requirements. 

Their goal is to enable efficient third-party auditing of AI applications, to lower the barriers to understanding, and contextually trusting models.

This UNIDIR Primer provides an overview of the main opportunities and limitations of synthetic data in the training of AI systems. While synthetic data can be a proxy for real-world data and help shorten training cycles, among other benefits, there are also significant risks and challenges associated with its use.

The State of Digital Health Report 2023 offers a comprehensive analysis of global digital health trends based on data from 67 countries across all WHO regions participating in the Global Digital Health Monitor (GDHM), complemented by secondary data from other sources. The GDHM, a multi-stakeholder, web-based platform, provides insights into the digital health maturity of countries across seven digital health enabling environment component areas. This report reflects progress since the State of Digital Health Report 2019, expanding on the GDHM’s participation from 22 countries in 2018 to 67 in 2023. It also contextualizes the analyses within a broader context of complementary trends analyses and digital health initiatives. The year 2023 witnessed tremendous activity worldwide in the digital transformation of health systems, culminating in the launch of the World Health Organization’s Global Initiative on Digital Health.

Following 3-day ‘marathon’ talks, the Council presidency and the European Parliament’s negotiators have reached a provisional agreement on the proposal on harmonised rules on artificial intelligence (AI), the so-called artificial intelligence act. The draft regulation aims to ensure that AI systems placed on the European market and used in the EU are safe and respect fundamental rights and EU values. This landmark proposal also aims to stimulate investment and innovation on AI in Europe.

La registrazione del keynote "Antropocene e distruzione delle value chains" del Dott. Marco Manca ai GreenBlueDays (Napoli 2023)

Plot twist? Human radiologists beat ML tools for radiology ^_^

The healthcare industry has been seeking the holy grail of patient engagement for decades and yet, here we have a prime example of successful patient engagement that's growing at scale and we are being "reminded to not overuse it". [Head. Desk.]  We need to seize the opportunity to incorporate smarter ways of supporting patient engagement and patient autonomy by way of:

👉 telehealth and virtual care

👉 peer health support communities

👉 digital heath tools

👉 RPM

👉 patient-reported outcomes

We need policy levers and reimbursement frameworks to support access to the above. 

Digital transformation has become a buzzword that is permeating multiple fields, including public administration and management. However, it is unclear what is transformational and how incremental and transformational change processes are linked

Clinical prediction models estimate an individual's risk of a particular health outcome. A developed model is a consequence of the development dataset and model-building strategy, including the sample size, number of predictors, and analysis method (e.g., regression or machine learning). We raise the concern that many models are developed using small datasets that lead to instability in the model and its predictions (estimated risks). We define four levels of model stability in estimated risks moving from the overall mean to the individual level. Through simulation and case studies of statistical and machine learning approaches, we show instability in a model's estimated risks is often considerable, and ultimately manifests itself as miscalibration of predictions in new data

Growing life expectancy poses important societal challenges, placing an increasing burden on ever more strained health systems. Digital technologies offer tremendous potential for shifting from traditional medical routines to remote medicine and transforming our ability to manage health and independence in aging populations. In this Perspective, we summarize the current progress toward, and challenges and future opportunities of, harnessing digital technologies for effective geriatric care. Special attention is given to the role of wearables in assisting older adults to monitor their health and maintain independence at home. Challenges to the widespread future use of digital technologies in this population will be discussed, along with a vision for how such technologies will shape the future of healthy aging.

Tempi irrealizzabili, carenza di risorse qualificate, vincoli dettati dal PNRR, molti progetti non saranno conclusi nei tempi previsti. Cosa succederà?

Leggendo i capitolati tecnici degli appalti specifici del PNRR relativi ai progetti di sanità digitale, tra cui quelli relativi alle cartelle cliniche elettroniche (CCE), mi è venuto in mente il mio insegnante di italiano quando ci spiegava le diverse forme del periodo ipotetico, in particolare la terza (quella dell’irrealtà).

I dati forniti dal rapporto Openpolis e Forum del Terzo Settore raccontano di una realtà tanto triste quanto prevedibile: il Terzo Settore, ma più in generale i temi dell’inclusione sociale e della lotta alle disuguaglianze hanno faticato moltissimo a trovare posto nel Pnrr e forse mai l’avranno. La previsione pessimistica era purtroppo facilissima da fare. Il Pnrr, figlio di due Governi diversi ma egualmente distanti dall’immaginare un ruolo di protagonismo del Terzo Settore, nasce culturalmente schiacciato tra due concezioni: la superficiale riverenza alle istituzioni europee con il quale il Governo giallorosso ha collassato qualunque riflessione sulla ripetizione ossessiva del mantra green/digitale del Recovery Plan da una parte e l’efficientismo da grande società di consulenza anni Novanta che incarnava buona parte della compagine del Governo Draghi dall’altra.

Understanding how humans should and do control artificial intelligence (AI) systems is central to many research areas, with applications ranging from self-driving vehicles to cybersecurity and national defence. We analyse a multi-disciplinary body of literature to show gaps and inconsistencies in existing approaches to human control of AI, highlighting, in particular, the practical challenges that stem from supervisory, task-specific, human control as a prominent paradigm. We conclude our analysis with a proposal to move away from this paradigm and instead consider approaches based on cooperating agents and the human–machine teaming (HMT) paradigm which, we argue, fit better with the capabilities and risks posed by AI systems.

In 2021, it was found that 79% of the Dutch population between the ages of 16 to 74 possess basic or above-basic digital skills. 📈 This puts the Netherlands in close proximity to achieving the European target of 80% digital proficiency by 2030. However, despite this progress, the prevalence of misinformation in social media posts about health remains a challenge. Therefore, our Dutch partners stress the importance of enhancing eHealth skills, particularly among individuals aged 50 and above.

GPT-3.5 and GPT-4 are the two most widely used large language model (LLM) services. However, when and how these models are updated over time is opaque. Here, we evaluate the March 2023 and June 2023 versions of GPT-3.5 and GPT-4 on four diverse tasks: 1) solving math problems, 2) answering sensitive/dangerous questions, 3) generating code and 4) visual reasoning. We find that the performance and behavior of both GPT-3.5 and GPT-4 can vary greatly over time. For example, GPT-4 (March 2023) was very good at identifying prime numbers (accuracy 97.6%) but GPT-4 (June 2023) was very poor on these same questions (accuracy 2.4%). Interestingly GPT-3.5 (June 2023) was much better than GPT-3.5 (March 2023) in this task. GPT-4 was less willing to answer sensitive questions in June than in March, and both GPT-4 and GPT-3.5 had more formatting mistakes in code generation in June than in March. Overall, our findings shows that the behavior of the same LLM service can change substantially in a relatively short amount of time, highlighting the need for continuous monitoring of LLM quality

Innovaties zoals intelligente chatbots en gebrekkig inzicht in bestaande algoritmes zijn op dit moment de belangrijkste algoritmerisico’s. Dit blijkt uit de eerste Rapportage Algoritmerisico’s Nederland, die de nieuwe directie Coördinatie Algoritmes van de Autoriteit Persoonsgegevens (AP) vandaag heeft gepubliceerd.

It is important that our health system can utilise available data when determining where to place healthcare resources. The authors aimed to demonstrate the feasibility of developing an interactive application that makes geographical and demographic data accessible to lay audiences, to develop a clinic planning tool that allows users to compare patient accessibility under hypothetical new clinic locations

An academic survey of patients' attitudes towards AI opening with the sentence "Artificial intelligence (AI) has the potential to improve diagnostic accuracy. Yet people are often reluctant to trust automated systems, and some patient populations may be particularly distrusting." immediately gives off the bias in the report... it's however a reference that we believe should be kept at hand, albeit handled with the appropriate correcting weights concerning biases.

The following topics are under consideration for inclusion in the next IHI call for proposals, which we plan to launch in summer 2023. IHI is publishing the draft topic texts in advance of the official call launch to give potential applicants additional time to start building a consortium and drafting a proposal.

La Regione Sardegna si prepara per partecipare alle gare delle Regioni capofila per l'attivazione dei servizi di telemedicina (che sono Lombardia e Puglia), così come disciplina il DM del 30 settembre 2022, attraverso le risorse del PNRR dedicate che saranno ad essa assegnate

Se c’è una cosa che ci sta insegnando il PNRR Digitale è che dobbiamo essere sempre in allerta per le continue novità. Restare sempre aggiornati è una bella sfida, ma ci sono delle risorse online che ci possono aiutare. Ecco, quindi, i dieci siti che non puoi non conoscere se lavori su questi temi

The World Health Organization signed an agreement with HL7 to jointly develop guidelines and advance the adoption of open interoperability standards that WHO says are critical to the development of equitable and evidence-based digital health.

After several desperate weeks, I decided to send a portal message to my entire panel of patients. I was uncertain how they would respond. In this letter, I confessed to feeling burned out, expressed my desire to continue to provide attentive care, then laid out several guidelines for portal use. The first was that I was going to stay off the EHR on weekends, and that if there was an emergency, telephone coverage with an on-call physician was always available.

The survey paper on Evaluations of LLMs now has a GitHub repo with the entire paper list. Bookmark this one!

The Institute for Technology, Ethics and Culture (ITEC), housed at the Markkula Center for Applied Ethics, is a collaboration between the Center and the Vatican’s Dicastery for Culture and Education. "Ethics in the Age of Disruptive Technologies: An Operational Roadmap” or, more briefly, the “ITEC Handbook,” offers organizations a strategic plan to enhance ethical management practices, empowering them to navigate the complex landscape of disruptive technologies such as AI, machine learning, encryption, tracking, and others while upholding strong ethical standards.

This publication presents the findings of a global horizon scan of innovations in science and technology that could help solve global health challenges.  An expert group scored over 100 innovations for their potential impact and the chance of wide adoption within 5,  5–10 or > 10 years. They also discussed enablers that would facilitate adoption of the innovations.

Machine-learning methods for analyzing genomic, transcriptomic, epigenomic, proteomic, and metabolomic data sets have yielded clinically directive information, mostly for rare genetic diseases.

Call for abstracts extended to July 31st 2023 - CompBioMed Conference 2023 (CBMC23) will address all aspects of the rapidly burgeoning domain of computational biomedicine, from genome through organ to whole human and population levels, embracing data driven, mechanistic modelling and simulation, machine learning and combinations thereof. We welcome contributions from academic, clinical and industrial participants alike.

Position paper with focus on assessing the status quo and proposing an action plan for establishing Public and Patient Involvement (PPI) within in silico medicine. While PPI has gained significant momentum in the wider biomedical community, it is currently underserved in the in silico medicine domain. The task force therefore aims to create awareness, support implementation, and enable a culture change within the in silico medicine community to maximise societal impact together with patients and the public

Computational models in the medical field are more and more stepping out the door of the laboratories they were developed in to find practical applications in the clinical or regulatory practice, for instance. In fact, terms such as digital twin in healthcare or in silico trials sound now very familiar to the scientific community. Still, a lot is about trust, and before a model is adopted by a clinician, a manufacturer, a regulator, its credibility must be demonstrated. In 2018 ASME issued the V&V-40 technical standard, which provides a Context of Use and Risk-oriented framework to establish the credibility of computational models. Alessandra Aldieri and Cristina Curreli will give an overview of this framework highlighting its key steps and provide our experience about its application.

The European Commission’s plan to reshuffle cohesion funds to finance innovative technologies raises concerns over the repeated use of the EU’s place-based policy to address new emerging needs. 

-not all regions can engage in the same way on deep tech innovation investments-

Call for papers!!!

The recent advancements and developments reached in Cardiovascular MedTech, especially over the past year, have been exceptional and are the result of the great achievements made by researchers and scientists thus far.

In order to explore this fast-growing field, Frontiers is launching a new series of Research Topics focusing on the evolution of methods, processes, techniques and ways of thinking in the Cardiovascular field that lead to new discoveries and research approaches. In this article collection, we seek contributions exploring the changing context and the rising new perspectives within Cardiovascular MedTech.

Congratulazioni agli amici di Cardioonline!!!
Il monitoraggio cardiaco degli astronauti di Virgin Galactic è fatto dai loro esperti ^_^

E' stato istituito dal Ministero della Salute il Tavolo Tecnico per lo studio delle criticità emergenti dall'attuazione del Regolamento dell'assistenza ospedaliera e dell'attuazione del Regolamento dell'assistenza territoriale.

Ebbene chi partecipa?

#18 rappresentanti - a vario titolo - della sanità, tutti esclusivamente #uomini (vedi immagine).

Provo a taggare chi di loro è qui su LinkedIN:

Marco Mattei, Domenico Mantoan, Francesco Enrichens, Anselmo Campagna, Francesco Saverio Mennini,PhD, Fabio De Iaco, Diego Foschi, dario manfellotto, Magi Antonio, Americo Cicchetti, Vito D'Andrea, Silvestro Scotti (mi scuso per eventuali imprecisioni).

Come scrive Simonetta Molinaro: "[...] Credo che debbano essere gli esperti chiamati al tavolo a chiedere che venga rimodulato in un'ottica di parità di genere. Il problema non sarà certo trovare professioniste competenti [...]"

Affermazione avallata da un notevole numero di "esperte in sanità" e, in primis, da Donne Protagoniste in Sanità | Community della D.ssa Monica Calamai.

Sul web non trovo articoli a denunciare questa situazione di fatto.

Solo l'ex ministra #Beatrice #Lorenzin ha rilasciato una dichiarazione:

"Nulla di personale nei confronti dei componenti del Tavolo istituito dal Ministero della Salute per la revisione degli standard di ospedali e territorio, ma possibile che in tutta Italia non ci sia neanche una donna che possa farne parte? DG, presidenti di società scientifiche, accademiche, medici del territorio, esperte della materia. Proprio neanche una?

Questo lungo elenco di nomi solo maschili stride proprio con la fotografia del mondo salute: in maggioranza netta fatto dalle donne. Tutto questo è umiliante per le tantissime professioniste del SSN. Spero che il governo cambi metodo mostrando volontà e maggiore sensibilità nell'indicare anche le donne nei tavoli decisionali. Il mondo salute è un mondo sempre più al femminile ed è strabiliante che poi il tavoli del “potere” siano tutti maschili!".

This paper presents an OMI risk score, developed by analysing consecutive patients with chest pain recruited from multiple clinical sites in the United States wih AI. The risk score provided superior rule-in and rule-out accuracy for OMI, boosting the sensitivity by ~28 percentage points and the precision by ~32 percentage points compared to reference standards. When combined with the judgment of experienced emergency personnel, the presented OMI risk score helped correctly reclassify one in three patients with chest pain.

The study sought to understand the potential roles of a future artificial intelligence (AI) documentation assistant in primary care consultations and to identify implications for doctors, patients, healthcare system, and technology design from the perspective of general practitioners. Three main themes emerged: professional autonomy, human-AI collaboration, and new models of care. Major implications identified within these themes included (1) concerns with medico-legal aspects arising from constant recording and accessibility of full consultation records, (2) future consultations taking place out of the exam rooms in a distributed system involving empowered patients, (3) human conversation and empathy remaining the core tasks of doctors in any future AI-enabled consultations, and (4) questioning the current focus of AI initiatives on improved efficiency as opposed to patient care.

This study was prepared by the European Parliament’s Policy Department for Citizens’ Rights and Constitutional Affairs at the request of the JURI Committee. Commercial, industrial and military applications of metaverse bring both opportunities as well as significant concerns for everyday life, health, work, and security. Legislative initiatives promoting fundamental principles of law, legislative and judicial oversight, applied comprehensively across a broad range of policies, are necessary to make sure that metaverse will play a positive role.

The text contains measures to prevent abuse of contractual imbalances in data sharing contracts due to unfair contractual terms imposed by a party with significantly stronger bargaining position.

Moreover, the text provides additional guidance regarding the reasonable compensation of businesses for making the data available, as well as adequate dispute settlement mechanisms.

This report distils technical and organisational lessons learned from the scientific work of the JRC that can inform the scoping and implementation of common European data spaces as envisioned by the European Strategy for data. Those lessons stem from JRC’s long term commitment and relevant research. The audience targeted by this report comprises policy officers and domain experts that are currently working on or expect to work on the scoping and implementation of common European data spaces. The report introduces the policy context, and positions JRC’s evidence as an instrument for translating policy conceptualisations and theoretical understandings of data spaces into a practical framework for implementation. This is followed by a synthesis of the requirements for common European data spaces derived from relevant policy documents. A set of data sharing how-to guides is provided in addition to a description of a dedicated knowledge base that corresponds to the functional and non-functional requirements of data spaces. In the investigation of the technical and governance aspects of data spaces, there is no single approach that can be applied for their setup. Therefore, a community-based approach through co-creation of data spaces that considers the domain-specific context is the only feasible way forward that would ensure buy-in by a broad spectrum of stakeholders. Finally, scientific evidence should continue to play a central role in the conceptualisation and implementation of European data spaces.

Just published: a systematic review of cardiac in-silico clinical trials. We definitely have some homework to do...
In the end 36 papers were included: 

- the first thing to notice: one must differentiate between studies with the population being digital twins, population models, or hybrid. Important difference in population size.

- One of the main red flags seen across literature is the lack of demographics reported. Maybe because we still see in-silico clinical trials through the engineering lens rather than as translational research? But the clinical application claims were sure common and heavy!

- One of the saddest red flags is that after listening to and reading many discussions on "generating trust in models" to accelerate innovation uptake, most papers to not share validations and uncertainty quantification.

Eugenio Luciani ha condiviso con noi le diapositive della sua presentazione al Customer eXperience Now 2023 - patient experience e fidatezza nel branding in digital health

SCImPULSE Foundation condivide la trascrizione della presentazione del Presidente Dr Marco Manca dal titolo "Il futuro della medicina di genere, all'incontro tra nuove tecnologie dell'informazione e una società che matura"

 Tobias Duswald (together with co-authors Ernesto A.B.F. Lima, J. Tinsley Oden, Barbara Wohlmuth) has recently shared a very exciting pre-print going by the title "Bridging Scales: a Hybrid Model to Simulate Vascular Tumor Growth and Treatment Response"

If managers assume a normal or near-normal distribution of Information Technology (IT) project cost overruns, as is common, and cost overruns can be shown to follow a power-law distribution, managers may be unwittingly exposing their organizations to extreme risk by severely underestimating the probability of large cost overruns. In this research, we collect and analyze a large sample comprised of 5,392 IT projects to empirically examine the probability distribution of IT project cost overruns. Further, we propose and examine a mechanism that can explain such a distribution. Our results reveal that IT projects are far riskier in terms of cost than normally assumed by decision makers and scholars. Specifically, we found that IT project cost overruns follow a power-law distribution in which there are a large number of projects with relatively small overruns and a fat tail that includes a smaller number of projects with extreme overruns. A possible generative mechanism for the identified power-law distribution is found in interdependencies among technological components in IT systems. We propose and demonstrate, through computer simulation, that a problem in a single technological component can lead to chain reactions in which other interdependent components are affected, causing substantial overruns. What the power law tells us is that extreme IT project cost overruns will occur and that the prevalence of these will be grossly underestimated if managers assume that overruns follow a normal or near-normal distribution. This underscores the importance of realistically assessing and mitigating the cost risk of new IT projects up front. 

DiMe proposes Evidence DEFINED, their approach to define ‘what good looks like’ when it comes to evidence evaluation for digital health products

Alla luce della crescente importanza della Modellazione e Simulazione (M&S), appare fondamentale una migliore diffusione delle pratiche e degli strumenti di M&S. La modellazione e la simulazione sono molto apprezzate nella scoperta scientifica perché forniscono approfondimenti aggiuntivi che spesso sono poco pratici da cogliere attraverso la sola analisi sperimentale e teorica del mondo reale. Una chiara comprensione dei vantaggi che queste tecnologie apportano all'ecosistema della ricerca aiuterà gli scienziati a creare valore aggiunto per le aziende che ricorrono a una ricerca e sviluppo basata su software.

L'M&S computazionale è una risorsa inestimabile nel settore sanitario. Tuttavia, la necessità di know-how avanzato e risorse computazionali ne limita l'adozione principalmente alle grandi aziende biotecnologiche e farmaceutiche. Rendere M&S disponibile a un ampio spettro di potenziali utenti (dispositivi medici e aziende farmaceutiche, ospedali, istituzioni sanitarie) richiederebbe un accesso facile e controllato alle risorse M&S in un ambiente sicuro.

Molti scienziati parlano della necessità di Good Simulation Practices (GSP), come regole per la modellazione e la simulazione applicate alle industrie farmaceutiche e dei dispositivi medici. C'è un forte bisogno oggi di avere regole chiaramente definite e regolamenti accettati; Gli SPG acquisiranno presto un ruolo essenziale. Nei prossimi anni la ricerca scientifica sulla medicina in silico dovrà diventare sempre meno speculativa, e sempre più traslazionale, cioè immediatamente applicabile e utile per applicazioni cliniche e industriali.

Infine, l'identificazione di buone pratiche di simulazione è particolarmente importante per la competitività dell'ecosistema delle scienze della vita in Europa, per evitare che l'industria negli Stati Uniti e in altre parti del mondo abbia un vantaggio competitivo significativo che renda più veloci le sperimentazioni cliniche.

Did Zheng Yongping, just admit, that the greater the humanization of surveillance technology, the more likely it is to be adopted??? “Elderly people who are in good physical shape are reluctant to install cameras in their homes, seeing it as an invasion of privacy, but if an intelligent robot accompanies them like a family member, they won’t feel so lonely if they live alone.” 

Latest overblown AI claim: GPT-4 achieves a perfect score on questions representing the entire MIT Math, EE, and CS curriculum.

The paper, released on arXiv two days ago, is getting a lot of attention (including on LinkedIn).

But some issues with it were immediately apparent:

➤ Answers were graded automatically by GPT-4 itself.

➤ When the LLM gave a wrong answer, it was given multiple chances to try again with different prompt formats until the correct answer was achieved.

The researchers didn’t make their data publicly available to prevent contamination of future LLMs, but three MIT undergrads managed to dig the test set out of the authors' git history. This led to some interesting discoveries:

➤ 4% of the test set questions were unanswerable. Some of them weren’t even questions. How did GPT-4 get them right?

➤ 5% of the test set questions were duplicates.

➤ Many of the few-shot prompts provided to the model contained the answer, largely due to leakage of information between multi-part questions.

The incisive critique by MIT students Raunak Chowdhuri, Neil Deshmukh, and David Koplow is well worth reading. 

DiMe’s next project, International Regulatory Pathways for Digital Health, will build on the success of our Digital Health Regulatory Pathways project to support digital health product developers to pursue and optimize their regulatory, evidence generation, and business strategies across multiple global markets and regulatory regions.

Together with our project partners, we will combine the expertise and experience of the project team with current global digital health product regulations, regulatory science, and industry experience to create a set of regulatory pathway maps for key global markets and analyses supporting multi-market evidence generation strategies, supplemented by case examples of digital health products successfully achieving multi-market access through multi-market evidence plans along with implementation toolkits.

If your organization is currently considering or actively deploying digital health products across global markets, you're invited to be share your interest in joining this work through a funding commitment and participation on the project team. Once you complete this form, we will reach out about next steps.

Il Laboratorio Sanità 20/30 torna a Napoli, il 6 Luglio 2023. Una tappa importante verso il 18° Forum Risk Management in Sanità. Una importante occasione di confronto sulle opportunità offerte dall'innovazione digitale e biomedica per il rilancio della sanità in Campania e nelle altre regioni del Sud.

L’obiettivo è un confronto tra mondo della sanità, mondo della ricerca e competenze tecnologiche per monitorare come può cambiare la sanità e come questa può essere cantiere per lo sviluppo della ricerca e dell’innovazione.

Il ruolo centrale spetta alle Aziende Sanitarie in rete con le università e i centri di ricerca e parchi tecnologici, per ottimizzare le opportunità di innovazione nella governance delle risorse umane e nei percorsi clinici e assistenziali.

Il “Laboratorio” è promosso da Gutenberg, Fondazione per l’Innovazione e la Sicurezza in Sanità, con il patrocinio di Regione Campania, Conferenza delle Regioni e PA, Comune di Napoli, Age.na.s e Istituto Superiore di Sanità.

Lori Wallach, Director Rethink Trade at American Economic Liberties Project, describes how Big Tech lobbyists are using international trade law to try to PRE-EMPT AND BLOCK everything we are working for in AI oversight and privacy.

The same is happening in Europe. What we can do: spread the word to colleagues in the tech world that there is a trojan horse sneak attack in this field and remember that IF it has some ground (most Countries have legiferated to solidify their obligation in the face of international threaties) the described logic of lobbying is essentially a "Dracula strategy" of organizing -- you must drag it into the light for public opinion to hold companies and politicians accountable! Write about it, organize peer letters (those have impact), mention the issue id your lectures and interviews, and teach about this to students!!!

Please note the use of the word SEARCH in these declarations by Google... we are in front of evasive medical entrepreneurship strategies at their best. When does this activity become illegitimate medical advice? Does the company get to decide this?

“Describing an odd mole or rash on your skin can be hard to do with words alone. Fortunately, there’s a new way Lens can help, with the ability to search skin conditions that are visually similar to what you see on your skin,” Google says.

“Just take a picture or upload a photo through Lens, and you’ll find visual matches to inform your search. This feature also works if you’re not sure how to describe something else on your body, like a bump on your lip, a line on your nails or hair loss on your head.”

Contrary to most hype seeking alarmist warnings, AI isn't erasing human expertise... in facts, in radiology (by most indicated as one of the specialties most amenable to AI disruption) job postings are growing over time.

Read this interesting testimony about why most living-room AI theoreticians are overselling their products by selling an overly simplistic view of the healthcare value chain that they reduce to "tasks" to easily interface and connect... and why policy makers SHOULD ALWAYS interact with real domain experts and practitioners before buying into any "solution".

Particularly concerning (almost incomprehensible) is the suggestion that the relatively light-weight regulatory burden proposed in the AI Act, which should help ensure due diligence, might only apply to sufficiently risky (“significant risk of harm”) high-risk systems. This almost makes a joke of the long years of effort to ensure that all AI in the EU is responsibly deployed.

This is one of the reasons why AI is unlikely to drive a new generation of rapid technical innovation—there simply isn’t enough money in it. Again, Perez doesn’t quite put it like this, but at the start of each of her five surges there is a significant innovation, sometimes only seen retrospectively, that creates a new form of abundance through radical cost reduction, which then opens up a transformative new market. It’s this that attracts the finance capital in. (Think: Crompton’s Spinning Jenny, or Ford’s assembly line, or the invention of the microprocessor.)

Twitter thread about the recent announcement by Google about the general availability of VertexAI 

Digital health is much more than just what meets the eye. Listen to the young Italian researcher Nicolò Cogno talking about efforts to build biologically realistic and clinically relevant in silico models to inform optimal choices in radiotherapy, using BioDynaMo (the agent based simulation software developed and distributed by our partner project BioDynaMo) 

Collection of resource to learn about penetration testing and security with a hands-on approach - Because in digital health there's more than just "data & privacy".

Our President, Dr Marco Manca, will join the conversation on the first day (June 22nd, 14.00-15.30) and reflect on the potential role of clinical research and technological R&D in producing the means necessary to rip the veil instanced in current practices over gender distinctiveness, blinding medicine to what makes everyone of us so exquisitely unique and precious.

The authors hope to reduce the barrier to using self-supervised representations for developing medical imaging AI by open-sourcing our code as well as relevant technical details to allow the wider community to replicate and build on our study.

This study was conducted using retrospective data and as such, further rigorous health economic studies are required to quantify the impact of our method, which would account for other hurdles and costs in real-world deployment and generalization of medical AI; for example integration into clinical workflows, infrastructure, and other IT considerations.

Umar's Doctoral project is focused on the intricate process of cortical development. Specifically, he is exploring how a layer of progenitor cells evolves into a layered cortex, by investigating the rules governing cells proliferation, differentiation, and migration. The simulation incorporates intracellular determinants and how they modify the response to external concentration gradients, echoing a biologically realistic process of human cortical development.

The target of the simulation is to mimic the developmental timeline of the mouse neocortex, whose development  hinges on a carefully coordinated series of cell proliferation, maturation, and migration events, processes that work in unison to create a diverse population of neuronal and non-neuronal cell types underlying most brain and mental function as we understand them today.

Lancée à l’automne 2021 dans le cadre de France 2030, la stratégie nationale d’accélération sur la santé numérique vise à encourager la transition vers une médecine plus préventive, prédictive et personnalisée et à favoriser l’émergence d’un solide écosystème de la santé numérique en France. « Le multi-échelle, c’est la capacité à prédire des choses, des données, des comportements, à plusieurs échelles de temps, c’est-à-dire sur des durées variables allant de l’ordre de la microseconde à plusieurs dizaines d’années, et d’espace, donc allant de la cellule à l’organe, et en prenant même en compte les habitudes de vie du patient et son environnement », explique Hugues Berry, adjoint au directeur scientifique d’Inria en charge du domaine « Santé et biologie numériques » et co-pilote du programme pour l’institut.

Much hype about this one... but hold your horses for: "To gauge the potential clinical impact, a group of six physicians performed a qualitative evaluation of 100 randomly sampled readmitted cases that were captured by NYUTron following the trial’s conclusion. Physician review suggested that SOME true positive predictions by NYUTron are clinically meaningful, preventable readmissions."

Are you curious about the harms AI is or is not causing in the world? Check out the AI Incident Database, which you can contribute to as well. 

In silico trials consist of the virtual testing of medical devices and medicinal products. They may help support the regulatory approval process of a medical product and, in general, the whole lifecycle of the medical product.

This “In-depth analysis of legal and ethical requirements” is part of the “Ethical and Legal Framework” of the In Silico World project, and assesses the core pieces of legislation and ethical principles identified in the “Legal and Ethical Inventory“.

We use machine learning as a tool to study decision making, focusing specifically on how physicians diagnose heart attack. An algorithmic model of a patient’s probability of heart attack allows us to identify cases where physicians' testing decisions deviate from predicted risk. We then use actual health outcomes to evaluate whether those deviations represent mistakes or physicians’ superior knowledge. This approach reveals two inefficiencies. Physicians overtest: predictably low-risk patients are tested, but do not benefit. At the same time, physicians undertest: predictably high-risk patients are left untested, and then go on to suffer adverse health events including death. 

Under the three-month pilot, telemedicine services are only offered when a doctor deems it necessary and safe for target patients. They include patients with chronic diseases who have done their first in-person visits; paediatric patients seeking follow-ups (only during holidays and at night); people living in islands and other remote areas; disabled persons; senior citizens over the age of 65; and people dealing with infectious diseases.

Check out the slides from "Scaling Radiology to Serve a Billion People" by Prodigi's @Bargava

In the last 6 months Germany, Netherlands, and Slovenia have published their new digitalization strategies. All three countries differ a lot in almost everything - size, population, % of GDP expenditure for healthcare, healthcare system structure, and digital maturity. So what are their key goals for the future digital transformation?  - One big change is expected by the ministries: hospitals must begin considering IT as an integral component of healthcare provision. Consequently, the costs associated with digital services and tools will need to be incorporated into the price of healthcare service provision.

The global governance of AI is getting more salient and more confusing. The authors look at ethical codes, industry governance, contracts/licensing, standards, int'l agreements and some domestic law. They then highlight that this global governance doesn't just respond to transboundary challenges but instead as a form of lobbying by both forestalling regulation and sigalling; as 'talking AI into being' by making it inevitable, by replacing roles of states, reshaping the role of organisations, and by marginalising global voices.

...and if you are interested, you should read this one about AI supply chain https://arxiv.org/abs/2304.14749v2

"Enormous quantities of high-quality research on how people perceive and experience frontline public services, especially digital processes. Yet, only some of this research has ever been published"

Large language models (LLMs) represent a major advance in artificial intelligence (AI) research. However, the widespread use of LLMs is also coupled with significant ethical and social challenges. Previous research has pointed towards auditing as a promising governance mechanism to help ensure that AI systems are designed and deployed in ways that are ethical, legal, and technically robust. However, existing auditing procedures fail to address the governance challenges posed by LLMs, which display emergent capabilities and are adaptable to a wide range of downstream tasks. In this article, we address that gap by outlining a novel blueprint for how to audit LLMs. Specifically, we propose a three-layered approach, whereby governance audits (of technology providers that design and disseminate LLMs), model audits (of LLMs after pre-training but prior to their release), and application audits (of applications based on LLMs) complement and inform each other. We show how audits, when conducted in a structured and coordinated manner on all three levels, can be a feasible and effective mechanism for identifying and managing some of the ethical and social risks posed by LLMs. However, it is important to remain realistic about what auditing can reasonably be expected to achieve. Therefore, we discuss the limitations not only of our three-layered approach but also of the prospect of auditing LLMs at all. Ultimately, this article seeks to expand the methodological toolkit available to technology providers and policymakers who wish to analyse and evaluate LLMs from technical, ethical, and legal perspectives.

Jess Morley's Guide to Thinking Critically About AI in Healthcare: a themed annotated bibliography with brief summaries for each theme

Some EU Countries (e.g. Finland) already have messaging app based triage and "first visit" to better meet the healthcare demand by their population. By relying of messaging, 1 doctor can respond to 50-ish patients requests per hour, instead of 4-5/h. The growing competence of LLMs is, through this lens, clearly appealing. The pivot point will be made by those who will study and deploy an appropriate hand-over procedure to leave the exchanges in the hands of human who wasn't yet involved, when the machine is beyond its ideal conditions of confidence... and when security at scale will be built in these chatty parrots we call AI, of course ;)

Qui sotto, opportunamente censurate, vedete solo una piccolissima parte delle directory della #Asl1 dell'Aquila rese disponibili nel #darkweb. Ed è un disastro. Per noi Utenti, per noi Cittadini, per i nostri dati personali. E' pubblicato di tutto, dai dati personali di medici e pazienti, a procedure di gestione, documenti, ordini e quant'altro. Non è tanto il fatto che siano stati violati. Nessuno è al sicuro. Ma non puoi salvare dati #sensibili (patologie - anche gravi, numeri di telefono, indirizzi, etc.) in CHIARO, in un file excel o word, senza nemmeno una minima protezione. Ovviamente tra tutti i dati ci sono anche le utenze (user name e password) di accesso al #cloud, salvate come pdf. A proposito della affermazione che continuo a sentire:"se ho il cloud non ho problemi". Continuiamo pure a parlare dei massimi sistemi, ma qui mancano proprio le basi della #alfabetizzazione #digitale. #Asl1 #Aquila #databreach #cloud #cyberrisk 

In silico medicine describes the application of computational modelling and simulation (CM&S) to the study, diagnosis, treatment or prevention of a disease. Tremendous research advances have been achieved to facilitate the use of CM&S in clinical applications. Nevertheless, the uptake of CM&S in clinical practice is not always timely and accurately reflected in the literature. A clear view on the current awareness, actual usage and opinions from the clinicians is needed to identify barriers and opportunities for the future of in silico medicine.

Undergraduate exposure to traditional neurosurgery/neuroanatomy labs is limited in Cameroon. Neurosurgery/neuroanatomy practical skills are gained essentially using non-practical means. Most students found the UpSurgeOn tool user-friendly, saw the need to incorporate it as part of their training, and perceived it to be essential in getting acquainted with neurosurgical skills.

Ziekenhuizen verkopen met regelmaat patiëntengegevens aan fabrikanten van medische hulpmiddelen. De fabrikanten kunnen die patiëntendata vervolgens gebruiken voor de marketing van hun producten of doorverkopen aan andere partijen, zonder medeweten van de patiënt. ‘Dit is een schending van het medisch beroepsgeheim.’

In public, hospitals rave about artificial intelligence. They trumpet the technology in press releases, plaster its use on billboards, and sprinkle AI into speeches touting its ability to detect diseases earlier and make health care faster, better, and cheaper.

But on the front lines, the hype is smashing into a starkly different reality.

Caregivers complain AI models are unreliable and of limited value. Tools designed to warn of impending illnesses are inconsistent and sometimes difficult to interpret. Even evaluating them for accuracy, and susceptibility to bias, is still an unsettled science.

Are database studies that are explicitly designed to emulate past and ongoing randomized clinical trials (RCTs) of medications able to generate similar causal conclusions? Selected database studies can complement RCT evidence to enhance understanding of how medications work in clinical practice. Emulation differences, chance, and residual confounding can contribute to divergence in results and are difficult to disentangle.

<<In this cross-sectional study, a chatbot generated quality and empathetic responses to patient questions posed in an online forum. Further exploration of this technology is warranted in clinical settings, such as using chatbot to draft responses that physicians could then edit. Randomized trials could assess further if using AI assistants might improve responses, lower clinician burnout, and improve patient outcomes.>>

The AI Act is a flagship legislative proposal to regulate Artificial Intelligence based on its potential to cause harm. The European Parliament is now inching toward formalising its position on the file, after EU lawmakers reached a political agreement on Thursday (27 April).

The text might still be subject to minor adjustments at the technical level ahead of a key committee vote scheduled on 11 May, but it is expected to go to a plenary vote in mid-June.

Microsoft and Epic Systems announced that they are bringing OpenAI's GPT-4 AI language model into health care for use in drafting message responses from health care workers to patients and for use in analyzing medical records while looking for trends.  Epic Systems is one of America's largest health care software companies. Its electronic health records (EHR) software (such as MyChart) is reportedly used in over 29 percent of acute hospitals in the United States, and over 305 million patients have an electronic record in Epic worldwide. Tangentially, Epic's history of using predictive algorithms in health care has attracted some criticism in the past.

Unfortunately this one is behind paywall, but it is very relevant, we recommend you reading it.

La raccolta di riscontri circa la qualità dei servizi offerti assume significati differenziati in ragione delle specificità settoriali. Un primo aspetto degno di nota concerne l’oggetto della valutazione: cosa si valuta di una prestazione medica o di una seduta di psicoterapia? Secondo molti medici che operano con le piattaforme che abbiamo studiato, le recensioni dei pazienti si basano principalmente sugli aspetti comunicativi ed emotivi che traspaiono durante una visita o un teleconsulto.

For 30 years, the Americans had a different protocol, the Europeans another one, the British their own protocol, and the Germans their own protocol. A few years ago, we decided to have a universal standard, and then we were happy and relaxed. What's the problem? The problem is that this universal standard provides almost zero information for testing cuffless devices. If you see cuffless devices tested using this universal standard that we love, it's not enough.

The second is that we need another standard. And we don't need another one; we need several, because this is not one technology. Rama has taught me that they work differently. They are different in their background technology, in the kind of devices, in the intended use, and whether they are wearable. We need to end up with a standard that works for every kind of cuffless device. We'll probably need some time to come up with something useful.

More than algorithms: an analysis of safety events involving ML-enabled medical devices reported to the FDA

Safety problems with ML devices involve more than algorithms, highlighting the need for a whole-of-system approach to safe implementation with a special focus on how users interact with devices.

EMA has published a report summarising the mid-term achievements of its Regulatory Science Strategy (RSS) to 2025. In the human domain, progress was made in several areas, including:

1.     fostering innovation in clinical trials;

2.     promoting use of high-quality, real-world data in decision making;

3.     reinforcing patient relevance in evidence generation;

4.     contributing to health technology assessment bodies’ (HTA) preparedness and downstream decision making for innovative medicines;

5.     supporting developments in precision medicine, biomarkers and ‘omics.

Here, the authors review recent applications of ML across the clinical oncology workflow. They review how these techniques are applied to medical imaging and to molecular data obtained from liquid and solid tumor biopsies for cancer diagnosis, prognosis, and treatment design. They discuss key considerations in developing ML for the distinct challenges posed by imaging and molecular data. Finally, they examine ML models approved for cancer-related patient usage by regulatory agencies and discuss approaches to improve the clinical usefulness of ML.

La presa in carico integrata del paziente con multi-morbilità cronica da parte di un team multi-professionale e multi-disciplinare è un obiettivo previsto dai più recenti atti di indirizzo in materia di gestione delle cronicità sia a livello nazionale che regionale, ma gli aspetti organizzativi per raggiungerlo non sono ad oggi pienamente definiti.

Il progetto JADECARE in Toscana ha delineato e sperimentato alcuni elementi di un modello per la presa in carico del paziente complesso con particolare riguardo alla messa a punto di un sistema di teleconsulto multiplo e sincrono tra i componenti del team multiprofessionale e multidisciplinare per la condivisone periodica del piano assistenziale individuale del paziente.

A competition to assess computational tools for accurate, robust and computationally efficient inference of unknown parameters in complex cardiovascular biophysical models from physiological data. Read more to participate 

Machine learning can easily produce false positives when the test set is wrongly used. Just et al in Nature HumBehav suggested that ML can identify suicidal ideation extremely well from fMRI and we were skeptical. Today retraction and our analysis of what went wrong came out.

So what went wrong? The authors apparently used the test data to select features. Obvious mistake. A reminder for everyone into ML: never use the test set for *anything* but testing. Only practical way to do so in medicine? Lock away the test set till algorithm is registered.

https://www.nature.com/articles/s41562-023-01560-6

Side note: it took 3 years to go through the process of demonstrating that the paper went wrong. Journals need procedures to accelerate this.

BTW, pay attention: Confound Removal in Machine Learning Leads to Leakage https://arxiv.org/abs/2210.09232 

App-a-geddon: there is an app for everything in #healthcare (or multiple) but very little coordination between them. Fragmentation is not a new issue in healthcare but you'd hope that #digital does something to relieve those issues instead of exacerbating them.

The article is the summary of a panel at ViVE on how to tackle the integration problem, ideally by positioning tools earlier in the #patientexperience and coordinate from there.

But they question for me still remains: why are we still more likely to launch a new app then coordinate existing tools? Is it because the landscape is changing so rapidly? The need to leave a mark and gather data? Something else?

"Nobody cares about their health more than the patient does. We need to make sure healthcare is easier and more convenient so people are actually able to get the care they need."

The smaller a device becomes, the easier to integrate into everyday life. That worked well for #technology in the past and might now come to #healthcare. Traditional blood pressure monitoring involves larger machines and a cumbersome process that is not realistic for everyday use at home.

So here a #healthtech solution: Sky Labs CART-I Plus ring now measures blood pressure on the finger, continuously. That provides longitudinal real-world data for physicians to analyze and use in the treatment recommendations. And it can combine it with other #wearables data to give a more holistic picture of the user.

"Worn on the finger, CART-I Plus uses PPG and sensor algorithms to measure and monitor blood pressure continuously around the clock. It provides a range of blood pressure data, including nighttime blood pressure, morning blood pressure, and blood pressure variability."

Approval in South Korea has been achieved, with FDA and EU MDR clearance expected over the next year as well.

The decree of the French fast track, PECAN (prise en charge anticipée des dispositifs médicaux numériques) has been published. 

The PECAN pathway will accelerate the reimbursement of innovative digital medical devices (with a therapeutic or tele-monitoring function) allowing a fastest access to patients. DMD manufacturers can start the process of registering to PECAN by getting the following assessments in parallel: 

- Evaluation by the CNEDiMTS (HAS) 

- tool: platform EVATECH - ~60 days 

The HAS also offers free private and confidential meetings to discuss the clinical trial protocol, and any questions you might have on the evaluation process.

 - Certification of interoperability and security delivered by the ANS (Digital Health Agency) 

- tool: platform Convergence - ~60 days 

The certification requirements for interoperability and security are written in compliance with the ISO 10781 standard. 

This standard defines, in the form of requirements, the minimum level of guarantees expected for which the ANS will issue a certificate of conformity for DMDs. 

Citizens CONDITIONALLY support the secondary use of health data as long as it matches their ethical values! 

Citizen Demands: 

I. The Data Relationship 

1. Being able to access information about the secondary use of health data, in an understandable way, allowing them to be more engaged 

2. Having access to their data and know how they are used for secondary purposes. However, they want to choose how and when they are informed about the uses of their data 

3. That their values should inform what is beneficial to individuals and what constitutes the common good 

4. That decision-making processes rely on a plurality of views and actors to increase their trustworthiness, as for them the latter depends on who is involved in these instances 

5. Being given the opportunity to be involved in the lifecycle of health data, as they need to be engaged on a continuous basis. Otherwise, their relationship with data custodians and users can deteriorate 

II. The Power Balance 

6. Being provided with the opportunity for meaningful and active decision-making in the secondary use of health data, as they value the ability to exercise control 

7. To ensure the protection of individuals' identity, which they perceive as one of the best ways to balance the harms and benefits of the secondary use of health data 

8. That data users' intentions should be transparent and in line with purposes citizens support, as they think some users might share their values more than others 

9. That accountability could be enhanced through transparent and stronger mechanisms 

10. To foster good IT solutions to protect their data, beyond having a strong legal framework in place 

III. A Citizen-powered Framework 

11. That stakeholders respect principles that align with citizens' ethical values 

12. Having a framework which facilitates the secondary use of health data for purposes and benefits that they support, while minimising the potential risks they identify. 

"A source familiar with the outage said the National Cyber Security Centre, the Cabinet Office and other government agencies had been alerted to the incident, given the group’s role in sensitive areas such as Royal Navy training centres and security at Ministry of Defence bases.

People at sites including critical national infrastructure have resorted to using radios, pens and paper, the source said.

Some employees still have access to computers and email, and the company said its investigation was in the early stages and that it was too soon to tell if the failure was caused by a cyber-attack."

After test-driving GPT-4 for 6 months in medicine A new book by Peter Lee, Carey Goldberg Isaac Kohane will be released as an e-book April 15th and as a paperback May 3rd. Eric Topol, MD - the author of Deep Medicine and multiple recent review papers on Artificial Intelligence - had the chance to read it already. The authors had 6 months to test drive GPT-4 before its release, specifically to consider its medical use cases, and put together their thoughts and experience to stimulate an important conversation in the medical community about the impact of AI.

The (honestly not so) hidden burden of digitalization on medical practice... GPs spend a lot of time patching up system failures. We investigated how unnecessary disruptions and system defects impact on their work.
Access the paper AND/OR the summary

Da oggi è ufficiale l'implementazione dei dataset ad alto valore (HVD) che ogni Stato Membro della Commissione Europea dovrà rendere disponibili in OpenData

Many scientists and politicians are calling for the establishment of a legal and ethical framework to avoid potentially detrimental impacts from the use of AI. This note gathers links to the recent publications and commentaries from many international think tanks on artificial Intelligence

Germany unveils a new strategy to “bring healthcare into the 21st century” ...and as usual it looks more like a list of desired outcomes than a strategy...

The model is tested on paywalled questions, which are less likely to have been part of the training set (although there is no information about this, of course... just some tentative evidence of lack of memorization). The report acknowledges the gap between benchmarking & clinical utility... but then again, it offers no testing on real-world tasks and no clinician was involved in the study nor as an author of the manuscript.

If you work (or plan to) with biomedical machine learning, this checklist for sharing your product/results might be handy 

EMA has launched a pilot to give scientific advice on the intended clinical development strategy and proposals for clinical investigation for certain high-risk medical devices (all class III devices and class IIb active devices intended to administer and/or remove medicinal product(s)). As of today, manufacturers can submit their letter of interest to be part of the pilot on scientific advice which will be provided by the medical device expert panels.

A computer, ChatGPT, has now successfully passed the United States Medical Licensure Exam (USMLE ) without any training. However, put real world symptoms in it’s prompt and it gives a very canned response with extensive hedging and qualifiers, as if reading directly from WebMD! How should physicians think about these tools, and what role does AI have in the evolving digital health space? - Classical example of "just because we can, it doesn't mean we should"

This specification sets out requirements for Application Programming Interfaces (APIs) in Dutch Healthcare. It has been developed as part of the Nictiz API strategy

Tra i 40enni, infatti, solo un accesso su 10 è appropriato, mentre la percentuale degli accessi appropriati sale al 40% tra gli anziani. Crolla dunque la diffusa percezione che ad affollare in modo ingiustificato i Pronto soccorso siano gli anziani.

Lo dimostra uno studio italiano condotto dalla Società Italiana di Gerontologia e Geriatria (Sigg) e dalla Società Italiana Geriatria Ospedale e Territorio (Sigot), che per la prima volta ha preso in esame oltre 20 milioni di accessi al Pronto Soccorso in Italia in ogni fascia d’età.

AgID e AGENAS hanno sottoscritto un accordo quadro per la realizzazione di appalti innovativi nel settore della sanità.

AGENAS è l’ente pubblico nazionale che svolge attività di ricerca e di supporto nei confronti del Ministro della salute, delle Regioni e delle Province Autonome di Trento e Bolzano. Con decreto legge 27 gennaio 2022 n. 4 assume anche il ruolo di Agenzia nazionale per la sanità digitale (ASD) con l’obiettivo di guidare e assicurare il potenziamento della digitalizzazione dei servizi e dei processi in sanità.

Da maggio 2024 scattano le sanzioni per i medici di medicina generale che facciano uso di cartelle cliniche elettroniche e sistemi di prescrizione non validati e certificati da enti terzi notificati, rispettando i criteri europei della normativa Ce-Mdr in classe 2x.

A government watchdog called for greater federal oversight of ethics boards that sign off on scientific studies, finding that for-profit companies have taken an outsize role in approving certain research and questioning whether financial motivations could put human subjects at risk.

L’autorità, nel parere espresso, chiede maggiori garanzie per la protezione dei dati.

Il parere è stato formulato su uno schema di decreto che definisce le modalità di accesso alle prescrizioni di medicinali rilasciate nel territorio italiano a pazienti che intendano utilizzarle in un altro Stato membro dell’Unione europea.

Il Garante, ricordando la necessità di assicurare una tutela rinforzata dei dati sulla salute, ha fornito precise indicazioni al Ministero della Salute per superare le diverse criticità presenti a suo giudizio nel testo.

-originally shared on LinkedIN by Suchi Saria-
Seeing ChatGPT and healthcare in the same sentence these days often scares me. Very often, the proclamation is about how ChatGPT will solve healthcare, starting with a small demo example and drawing far reaching conclusions from there. ChatGPT is a word predictor (given past words/context). It generates plausible ideas in a way that mimics human speech; so it sounds intelligent but is not designed to be factually grounded/correct. This article (on Cerner’s blog nonetheless) discusses using ChatGPT to create clinical vignettes (reasonable if verified by an expert) and potentially a diagnosis tool (bad plan b/c of how ChatGPT works!): https://lnkd.in/eNMwEBDK This is not to say that we haven’t made ground breaking progress in AI. We have. And we’ve even built amazing AI tech to serve as a clinical co-pilot / augmentation tool for clinicians to do all sorts of useful things including assist w/ diagnoses: see https://lnkd.in/e_Wkb9cS by Bayesian Health But doing so meant leveraging the AI ingredients to build recipes that prioritize veracity, trustworthiness, transparency, robustness, and remove human bias. More broadly, in a time of AI hype, enthusiastic declarations are being made by many in a rush to leverage PR to their advantage. When the PRs are coming from big companies, folks are even more wanting to go along. Highly speculative ideas confidently written w/o proof benefits no one: not the people building these systems nor their end users (in James Vincent’s words).

Automating adverse events reporting for pharma with Amazon Connect and Amazon Lex - Every pharmaceutical company manufacturing medicine must provide customers nationwide with a method to report adverse events following medicine usage as well as emergency assistance as needed. To comply with regulatory policy and enable an Adverse Events Reporting System (AERS), pharma companies must provide dedicated, toll-free phone numbers and contact center agents to handle inbound calls.

But they must also be prepared for sudden spikes in call volume, which can increase contact center agents’ workloads and lead to long wait times for customers. With these limitations comes the possibility that customers may not be able to report adverse events...

Lo scorso settembre il Garante della Privacy ha espresso parere negativo sullo schema di decreto che prevedeva la realizzazione della nuova banca dati denominata Ecosistema Dati Sanitari (EDS), prevista dalla riforma del Fascicolo sanitario elettronico. Il Garante ha chiesto anche di correggere un secondo schema di decreto, per favorire e migliorare l’implementazione a livello nazionale del Fascicolo sanitario elettronico (FSE).

La bocciatura del Garante, di cui potete trovare una spiegazione in questo post, è stata determinata dalla mancanza di un quadro normativo specifico per il nuovo FSE e per l’EDS. Quadro normativo che ancora oggi, a sei mesi dallo stop, non è stato ancora definito.

Revised Zero Draft [framework] Convention On Artificial Intelligence, Human Rights, Democracy And The Rule Of Law -  The Council of Europe is working on a Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law. Drafting is ongoing on what will be the first international convention on AI. The committee responsible for drafting the Convention has  produced a “zero” draft which is now being negotiated.

Download Medical Device Innovation Consortium (MDCI)'s new "Landscape Report & Industry Survey on the Use of Computational modeling & Simulation (CM&S) in Medical Device Development" available here: https://mdic.org/resource/cmslandscapereport/

We usually focus on tech and policy news... however, healthcare struggles with innovation, and resort to a neverending cycle of pilots that jeopardize care and citizens experience. Thus, we would like to start sharing a selection of reflections about innovation in practice, starting from this one: "Architecture modernization initiatives aim to convert aging architecture into modern architecture, applying the latest tech innovations and architectural patterns. To take full advantage of modernized architecture, it’s also necessary to modernize architectural practices and ways of working. When done right, this essential aspect enables modernization to be a continuous, value-driven activity in the organization, as opposed to complex multi-year initiatives with heavy up-front planning."

We are at a critical juncture in the advancement of telehealth.  We must demonstrate it is a care model on equal footing to office-based care and one that all parties in healthcare must take seriously. Telehealth has a trust problem. It’s real and we all own it. Hence, we all must do our part to help fix it. 

The purpose of the document is to provide a set of the specifications needed to develop an interoperable European Digital Identity (EUDI) Wallet Solution based on common standards and practices

What we say vs. what we do about privacy:

• 27% of people said they would hand over their unlocked phone to an experimenter to look at privately,  98% of people did 

• People give away their data for small payments as long as they can actively avoid learning what is done with it

FDA has now cleared more than 500 healthcare AI algorithms

DIME Society has a nice initiative around "measurements": pre-competitive research projects/consortia to re-define the measurements of personal health at the border between clinical and personal, by exploring standardization potential of new approaches embracing telemedicine.

Calling all European citizens! The PerMedCoE is conducting a survey on health app use and needs your input.
Help them conceive a quality label for apps that matter to you. 

One of the biggest questions with AI is how it interacts with skills. For example, adding robots to surgeries doesn't seem to do much for the best surgeons, but does it benefit bad ones? 

According to this report, low-skill surgeons are much worse than good ones, but robots close the gap by 50%! 

Despite a number of initiatives undertaken by the EU in the last few years towards advancing the development and uptake of AI technologies to help EU citizens better monitor their health, receive better diagnoses and more personalized treatments, as well as live a healthier and more independent life, current situation in the EU indicates that healthcare organisations are slow in implementing AI technologies in healthcare and that the level of adoption is low overall. To achieve its long-term objective of the effective implementation of AI in the healthcare sector the Commission plans to work on a common legislation and policy framework to yield the benefits that AI can bring. Based on evidence gathered in this study, while most EU MS that have developed AI strategies identify healthcare as a priority sector, there are no policies within those strategies targeting healthcare in particular. At the same time, EU MS have made progress in proposing regulatory frameworks around the management of health data which is a foundational element for the further development of AI technologies in the healthcare sector. In terms of adoption, while healthcare organisations in the EU are open to adopting AI applications, at present, adoption is still limited to specific departments, teams and application areas. The lack of trust in AI-driven decision support is hindering the wider adoption, while issues around integrating new technologies into current practice are also prominent challenges identified by relevant stakeholders in EU MS. The scientific output of EU MS in the area of AI in healthcare is largely attributed to the larger EU MS which are also the most active in collaborating between each other and with smaller MS. Additionally, a need is identified for further financial support to support the development of AI technologies which are translated into clinical practice, including support targeting the acquisition of Intellectual Property (IP) rights for the developed technologies. To promote the development and adoption of AI technologies in the healthcare sector, the Commission may address challenges related to policy supporting the further development and adoption of AI in healthcare, increase investment, enable the access, use and exchange of healthcare data, and develop initiatives to upskill healthcare professionals and to educate AI Developers on current clinical practices. Addressing culture issues around trust in the use of AI in the healthcare sector and creating or updating policy supporting the translation of research into clinical practice were also important insights extracted from this study.

This scoping review by WHO explores the extent of inequities in digital health across a comprehensive range of specific characteristics. It also identifies key areas that need to be considered when integrating digital health into health-care systems to ensure equity.

The Council of Europe has spearheaded demands for greater transparency and accountability for AI systems. Now, contrarily, it's developing an international convention on AI behind closed doors...
Healthcare allied professionals across EU are showing a range of sensitivities for the issue, the highest being in Germany... how can we inject the right information in mass communication to sensitize all medical professionals in every Country to a challenge that will hugely impact medicine and healthcare in the next future?

Helping medical doctors and patients in the Foundation Model A.I. era - Until now, the deep learning in healthcare has almost exclusively been unimodal, particularly emphasizing its applicability for all different types of medical images, from X-rays, CT and MRI scans, path slides, to skin lesions, retinal photos, and electrocardiograms.

In contrast, foundation models are multimodal, based upon large amounts of unlabeled, diverse data with self-supervised and transfer learning. These models are characterized by in-context learning: the ability to perform tasks for which they were never explicitly trained.

Recently, the University of Florida’s massive computing resource, GatorTron, developed from scratch a LLM using 8.9 billion parameters and >90 billion words of text from electronic health records to improve 5 clinical natural language processing tasks including medical question answering and medical relation extraction. While much smaller than Med-PALM’s model, this is one of the first medical foundation models to be developed by an academic health system, not large tech companies like Google, OpenAI, or Meta. 

It should be pointed out that it’s not exactly a clear or rapid path because there is a paucity of large or even massive medical datasets, and the computing power required to run these models is expensive and not widely available

Da lunedì 16 gennaio 2023 Ausl Ferrara attiva il nuovo servizio di Radiologia Domiciliare che si rivolge agli ospiti delle strutture sociosanitarie, Centri residenza anziani (Cra) e Centri socio-riabilitativi residenziali (Csrr) convenzionati, di tutto il territorio provinciale, nonchè ai pazienti ricoverati presso gli Hospice.

On March the 10-11 2023 the Italian Society of Telemedicine will host its International Bologna Consensus Assembly: don't miss out!!!
The scientific program, as of Jan the 9th 2023, is available here

"It’s actually the extreme exception to the rule if a clinical AI product has been submitted with data that backs up its effectiveness."

A look into participants' experiences with the SOAR knee program (a virtual physiotherapist-guided knee health program for persons at-risk of osteoarthritis after a sport-related knee injury).

Literature review on NLP in dentistry: literature is disconnected; the quality of reporting is getting better but doesn't quite match current standards in clinical NLP... we also need shared datasets