BODHI-S

BODHI-S is clinician curated clinical knowledge graph of conditions-symptoms-specialities.

About Dataset

BODHI-S is designed to map the complex web of patient presentations. It links clinical conditions (spanning disorders, lifestyle habits, allergies, and family/medical histories) to their associated symptoms, as well as the specific medical specialties equipped to treat them. Key Features of BODHI-S: Nuanced Symptom Mapping: The graph includes both root nodes (e.g., "Fever") and compound symptoms (e.g., "Fever with chills" or "Fever for 3 days"). Because the nuanced presentation of a symptom drastically alters the clinical probability of underlying conditions, treating these compound variants as distinct nodes allows for high-fidelity reasoning. Contextual Risk Factors: It maps condition-to-condition relationships, establishing vital prerequisites and risk factors. Demographic Probabilities: The graph embeds likelihood and triage categorizations for both symptoms and conditions, dynamically adjusted across age and gender cohorts. These probabilities are not arbitrary; they are derived from expert clinical consensus and augmented by normalized data across millions of Indian EHR records. Read more on BODHI: https://info.eka.care/services/bodhi-bharat-ontology-for-disease-healthcare-informatics Github: https://github.com/eka-care/BODHI?tab=readme-ov-file

Purpose of Dataset

1. Clinical Grounding For Generative Ai The Fundamental Risk Of Deploying Large Language Models (Llms) In Healthcare Is Confabulation—generating Plausible But Inaccurate Medical Outputs. Research Demonstrates That Llms Require Structured Grounding To Be Clinically Trusted. Graphrag Addresses This By Augmenting Ai With A Knowledge Graph, Allowing Models To Traverse Factual Relationships Rather Than Relying On Text Embeddings Alone. Bodhi Is Purpose-built For This Paradigm. Offering Over 13,000 Clinician-validated, Snomed-ct Linked Relationships, It Serves As The Ideal, Highly Accurate Foundation For Graphrag Pipelines Tailored To The Indian Healthcare Context. 2. Comprehensive Patient Health Profiling Capturing A Patient's Complete Health Picture Is Highly Valuable For Insurance Underwriters, Emr Platforms, And Preventive Care. Because Real-world Health Data Is Often Messy And Fragmented, Knowledge Graphs Act As A Unifying Framework. Through Reverse Inference, Implicit Data Becomes Explicit—for Example, Automatically Linking A Standalone Metformin Prescription To Type 2 Diabetes, Which Then Maps To Related Co-morbidities. Every Data Fragment, From Family History To Lifestyle Habits, Contributes To A Richer, Fully Connected Health Profile. 3. Symptom Checking, Triage, And Specialty Routing Frontline Health Workers Frequently Lack Advanced Diagnostic Tools. While Standard Ai Chatbots Carry Hallucination Risks And Require Internet Connectivity, Structured Clinical Knowledge Graphs Offer A Lean, Offline, And Deterministic Alternative For Resource-constrained Settings: Weighted Diagnoses: Matches Symptoms Against Statistically Weighted Probabilities To Accurately Rank Potential Conditions. Automated Triage: Instantly Categorizes Patient Presentations Into Emergency, Worrisome, Or Opd Managed. Specialty Routing: Automatically Directs The Patient To The Correct Medical Specialty Based On The Linked Condition.