Shantanu Seth

From Diagnosis Codes to Clinical Evidence: A Reusable AI Architecture for Finding Rare Disease Patients

Abstract:

Rare and complex diseases are often difficult to identify through traditional claims- based approaches because many lack definitive diagnosis codes, while critical clinical evidence remains buried in unstructured physician notes, laboratory results, and other fragmented data sources. This presentation introduces a reusable AI architecture for identifying rare disease patients by integrating unstructured EHR notes, laboratory and biomarker data, and structured claims into a unified patient-finding framework. The approach combines LLM/NLP-based extraction of hidden clinical signals, human-in-the-loop clinical validation, rigorous statistical and operational protocols, and machine-learning-based look-alike models to generate validated patient cohorts and prioritized HCP and account lists. The presentation also explores a governed GenAI architecture for clinical understanding and reasoning, evidence attribution, confidence scoring, and structured output, supported by privacy controls and continuous model monitoring. A production case study involving a rare endocrine disease demonstrates the practical impact of this methodology, achieving more than 96% LLM recall through chart-review validation while identifying hundreds of patients and previously untapped HCP opportunities for field engagement.

Profile:

Shantanu Seth is a management consultant and decision science leader with more than a decade of experience developing analytics, AI, and data science solutions for large global organizations, with a strong focus on healthcare and life sciences. He currently serves as Senior Director at Axtria, where he leads decision science engagements and advises biopharmaceutical clients on patient analytics, commercial model design, forecasting, segmentation, and ROI- driven optimization. Throughout his career, he has led cross-functional teams, scaled analytics capabilities, and delivered high-impact transformation programs across multiple geographies. Shantanu holds an MS in Business Analytics from the College of William & Mary and a BTech in Electronics and Communication Engineering.