Interactive Case Viewer and Portfolio Dashboard
The TA‑MedMal platform includes two interactive tools that demonstrate how structured concepts, keyword scoring and similarity analysis can be used to explore medico‑legal claims at both case and portfolio level. These tools are built on the same processing pipeline described above and provide transparent, reproducible insight into the underlying evidence.
The Single Case Viewer presents interactive medico‑legal reports for each of the 140 curated malpractice claims. Every case is processed through the full text‑analysis workflow, producing structured concepts, modified Tf‑Idf keyword scores, and similarity fingerprints.
Users select a case from a dropdown menu and are shown a structured, evidence‑based summary derived directly from the original letter of claim.
Page 1 – Claim Narrative and Concepts
The original narrative is displayed alongside extracted misadventure concepts (e.g., delayed diagnosis, surgical misadventure), ICD‑10 disease codes and OPCS‑4 procedure codes. The Viewer also highlights similar claims identified through concept fingerprints and keyword affinity, enabling rapid thematic comparison.
Page 2 – Tokenised Document Analysis
The claim is shown in tokenised form, with positions highlighted where misadventure, disease and treatment concepts are detected. Compound terms and proximity‑triggered concepts (e.g., “failure” near “diagnose”) illustrate how the system interprets the narrative.
Page 3 – Keyword Summary
This summary highlights the most informative clinico‑legal concepts in each case.
High‑weighted terms reflect concepts that the model identifies as especially important within the case, based on their clinical or legal relevance and how they appear in the text.
The model goes beyond simple word frequency. It groups related medical and legal expressions under shared canonical concept codes — for example, “breast cancer” and “invasive ductal carcinoma” are unified as C50 Cancer Breast.
This stabilises meaning, prevents duplication, and makes the colour‑coded visualisation easier to interpret.
By promoting clinically and legally significant terms, the viewer presents a clearer, more coherent picture of each case — supporting forensic review, risk management, and comparative analysis.
The Portfolio Viewer enables exploration of the same 140 claims at scale. Interactive dashboards allow users to examine misadventure types, disease codes, procedures, specialties and thematic patterns across the portfolio.
Page 1 – Claims Summary by Specialty
Bar charts show claim counts and costs by specialty. A filterable table allows users to drill down by misadventure type, ICD‑10 code, OPCS‑4 procedure and specialty group.
Page 2 – Keyword Theme Exploration
Treemaps display modified Tf‑Idf keywords for selected cohorts, with an optional severity‑weighted view. Filters allow exploration by specialty, misadventure, disease and treatment dimensions.
Page 3 – Misadventure Theme Analysis
Frequency and severity of misadventure categories (e.g., delayed diagnosis, postoperative complications) are shown using bar charts, line charts and heatmaps.
Page 4 – Claim Cluster Analysis
Anchor cases and ranked clusters are displayed with related cases and affinity scores. Heatmaps highlight shared keywords and concepts, while pie charts show contribution levels within clusters.
Page 5 – Explore Hospitals
This tab introduces institutional‑level exploration. Users can analyse claim patterns by hospital centre, comparing record counts and claim values across units. Dual heatmaps show TF‑IDF and claim‑weighted keyword saliency, enabling identification of dominant misadventure and disease themes within each centre.
Page 6 – Explore Clinicians
This tab drills down to individual practitioner profiles. Interactive charts display claim counts and costs per clinician, while treemaps reveal the language and procedural patterns associated with their cases. The view supports forensic analysis of recurring failure modes, procedural risks, and claim severity distributions.
The Portfolio Viewer supports semantic indexing, comparative analysis and pattern recognition across medical litigation data.
Why These Tools Matter
Together, the Single Case Viewer and Portfolio Viewer provide complementary perspectives:
Case‑level insight through detailed narrative analysis
Portfolio‑level intelligence through thematic and structural patterns
Entity‑level exploration through hospital and clinician dashboards
By combining concept extraction, keyword scoring and similarity mapping, the system supports:
identification of recurring patterns of medical error
benchmarking by specialty, misadventure type or procedure
underwriting and risk‑management decision‑making
evidence‑based governance and litigation strategy
These tools demonstrate how unstructured claim documents can be transformed into structured, auditable intelligence that supports both operational and strategic decision‑making.
MBB June 2026