Wadhwani AI (Lords Education And Health Society)

Global Standards, Local Impact: Democratization of SNOMED CT Standards in the world's largest democracy (India) by encouraging the adoption of standards among health providers of the national telemedicine service through an intuitive multilingual interface.

APAC

Artificial intelligence, Clinical Practice, Data quality, Mapping, Translation

India's national telemedicine service is globally the largest one with hub-and-spoke model of implementation. traditionally while hubs used SNOMED CT for diagnoses, spokes relied on error-prone free-text entries for symptoms. Despite the large-scale collection of patient data, the lack of standardization hindered its effective use for clinical decision-making and AI-driven solutions. To address this, a multilingual symptom repository was developed, translating 300 common symptoms and local synonyms into 12 Indian languages, each one mapped to appropriate SNOMED CT codes. The repository integrated SNOMED-linked multiple-choice questions (MCQs) for every symptom, capturing attributes like duration, location, and aggravating factors. Over 3000 terms‚symptoms, synonyms, questions, and answers‚were encoded using the SNOMED CT browser and validated by clinicians and public health experts, creating an intuitive, multilingual interface for health providers. Embedded in the telemedicine platform, this tool democratizes SNOMED CT: spoke providers now input data in local languages via dropdown menus populated with SNOMED-standardized terms, simplifying data entry. Patients benefit from precise documentation in their native language, reducing misinterpretations. Hub clinicians receive structured summaries to enhance diagnostic accuracy and treatment plans. Program managers leverage standardized datasets for real-time disease surveillance. Innovators gain machine-readable data to train AI-driven clinical support tools. This initiative breaks adoption barriers in low-resource settings by aligning SNOMED CT with linguistic diversity, fostering equity in digital health. Key outcomes include improved data quality, interoperability, and scalability for AI solutions. The project underscores the importance of collaborative validation, clinician engagement, and localized terminology mapping for SNOMED CT adoption.

Description

The project is geographically limited to India and precisely related to only the "chief complaint section" of case preparation under national telemedicine services. overall scope includes :

* Multilingual Support: Translating 300 common symptoms and local synonyms into 12 Indian languages, mapped to SNOMED CT codes, to address linguistic diversity.

* Structured Data Capture: Integrating SNOMED-linked multiple-choice questions (MCQs) to capture symptom attributes (e.g., duration, location) and reduce ambiguity.

* Clinician & Patient Empowerment: Providing spoke providers with intuitive, multilingual dropdown interfaces and enabling patients to receive care in their native language.

* Public Health & AI Readiness: Enabling real-time disease surveillance for program managers and generating AI-ready datasets for innovators.

* Scalability in Low-Resource Settings: Demonstrating SNOMED CT adoption in resource-constrained environments through collaborative validation and localized mapping.

Scope

The following are key reasons why we selected SNOMED CT for this intuitive multilingual interface.

1. Clinical Granularity
SNOMED Ensures precise symptom documentation to avoid misinterpretation between providers.
Captures "burning chest pain" (a subtype of "chest pain") vs. vague free-text entries.

2. Simplified Data Entry for Mid-Level Providers
SNOMED CT Minimizes training needs with intuitive, multilingual interfaces.
Example: Local terms like "‡§ñ‡§æ‡§Ç‡§∏‡•Ä" (Hindi for "cough") are easy to map to SNOMED codes via dropdowns.

3. Multilingual & Local Terminology Support
Mapping of SNOMED Codes to closely related local language terms bridges language barriers while maintaining standardization.
Example: 12 Indian languages mapped to SNOMED CT for seamless regional adoption.

4. AI-Ready Structure for Predictive Analytics
SNOMED CT Logic-based ontology enables symptom-pattern analysis.
Example: Attributes like "Cough duration: more than 15 days" train AI to infer conditions like Tuberculosis.

5. SNOMED CT was preferred over other standards because: ICD-11 is designed for billing/statistics, not real-time care. Lacks symptom granularity (e.g., no "cough with bloody sputum").
LOINC focuses on lab tests (e.g., "hemoglobin test"), not clinical narratives or symptoms, both standards provide limited multilingual flexibility and flat structures hinder AI training and frontline usability.

How SNOMED CT will be used

SNOMED CT Browser International Edition was extensively used at every step while ensuring global-level standardization for local terminologies.

1. Symptom-to-Code Mapping: 300 symptoms and English synonyms were translated into 12 local indic languages and mapped to SNOMED CT codes of standard symptoms using the SNOMED CT browser.
Ex: " Fever" is a standard symptom while local words for it in India are 1) Bukhaar 2) Jwaram etc. so all these terms were mapped to (Fever- finding, CTID: 386661006)

2. Structured Data Integration: SNOMED-linked MCQs captured symptom attributes (e.g., "duration"or "aggravating factors") as structured data, replacing free-text entries.
Ex: Local language question " Fever since how many days" was mapped to the SNOMED CT term "Duration (property) (qualifier value)SCTID: 762636008

3. Validation & Clinician Engagement: Over 3,000 terms (symptoms, questions, answers) were validated by clinicians and public health experts to ensure clinical accuracy.

4. Multilingual Interface Development: Dropdown menus in local languages were populated with SNOMED-standardized terms, enabling spokes to input data without coding expertise.

5. Hub-Spoke Workflow Enhancement:Structured summaries with SNOMED codes improved hub clinicians' diagnostic accuracy, while standardized datasets empowered program managers in surveillance.