Presenter Profile
Horst A. von Recum
Professor
Case Western Reserve University, Department of Biomedical Engineering
Professor Horst A. von Recum has been a faculty member in the department of Biomedical Engineering at Case Western Reserve University since 2004. He received his PhD in Bioengineering from the University of Utah in the laboratory of Prof. Sung Wan Kim in biomaterials for tissue engineering and drug delivery. His co-mentor was Prof. Teruo Okano from the Tokyo Women’s Medical University. Following his PhD studies he was a Postdoctoral Fellow at Harvard / MIT in the laboratory of Prof. Robert Langer, working on polymers for gene therapy, and implantable devices.
His research interests have been in the use of specifically designed polymer-drug interactions to control drug delivery, and allow for post-implantation refilling. Recently he has moved into using machine learning and big data models to explore unanticipated interactions between biomedical implants and unrelated therapeutic medications.
TALK TITLE
Using Machine Learning to Predict Unintended Drug-Implant Interactions
KEYWORDS
drugs, biomedical, materials, interactions, machine learning
ABSTRACT
Drug delivery is where implantable/attachable devices are used to administer therapeutic medicines for prolonged periods of time. Our lab pioneered the use of materials designed to have specific, tunable interactions with drugs. We found that materials with a high affinity for drugs can provide their long-term, sustained delivery, and those materials can even be refilled in vivo, for additional therapeutic windows of delivery. We have shown that such “refillable” devices can be used to deliver, and be refilled with, a range of drugs in several animal models; and for the first time be able to treat lifelong diseases and conditions.
Conventional polymers used in biomedical applications are not expected to refill after implantation since they do not have chemistries designed to promote polymer-drug interaction. Nevertheless, our laboratory discovered that some conventional polymers do refill with drugs following implantation, even without specifically designed interaction chemistries. This represents a huge, unanticipated risk; namely, with patients using drugs for conditions that are unrelated to their need for the medical implant (e.g. a cancer drug for a patient with an implanted cardiac stent).
To study this, our laboratory developed a machine learning model to predict unintended interactions between polymers and drugs. We populated this model with structure-function information from our designed polymers and were able to accurately predict polymer / drug interactions among the 2000+ drugs approved by the US Food and Drug Administration. Our next steps are twofold. Firstly, we are searching for available information on existing conventional polymers used in biomedical implants (“Material Genomics”). Secondly, where necessary we are using high-throughput methods to generate new data on existing polymers. With this model we hope to be able to explore drug/polymer interactions between existing implants and therapeutics; but also use it is as a tool in the development of new polymers and drugs.