NIME™ Symptom Checker Widget Architecture, Knowledge Base, Retrieval and Triage Logic, and Site-Wide Deployment in the NHOS Platform

Description The NIME™ Symptom Checker Widget is a browser-based symptom exploration and health-information retrieval component developed by NHOS Labs, Inc. as part of the NHOS health intelligence ecosystem. Built on the NHOS Intelligence Matrix Engine (NIME™), the widget provides a structured interface through which users can enter symptoms, explore potentially relevant condition records, review matched symptom information, and receive a heuristic suggested next step based on the system's rule-based safety and triage logic. Version 1.0.0 introduces a site-wide deployment architecture, a structured knowledge base of 713 condition records, pediatric-aware search behaviour, deterministic symptom matching and ranking, emergency-oriented phrase detection, and direct integration with the NHOS Health Encyclopedia and NHOS Children's Symptoms Checker. The widget is designed around local browser execution, explainable retrieval, conservative safety messaging, and privacy-conscious interaction. It does not use a trained statistical or machine-learning diagnostic model, does not calculate diagnostic probabilities, and does not transmit user-entered symptoms to a remote processing service. This technical report documents the widget's architecture, knowledge-base composition, retrieval and ranking methodology, safety and triage logic, privacy characteristics, integration interfaces, deployment approach, engineering verification, performance observations, and known limitations. 1. Knowledge Base and Content Coverage The documented NIME™ Symptom Checker knowledge-base snapshot, dated 31 October 2026, contains 713 unique condition records and 1,036 distinct symptom strings. The condition records are organized into age-relevance categories: 571 general, all-ages condition records. 55 adult-predominant condition records. 71 pediatric-specific condition records. 16 pediatric-relevant condition records. The combined pediatric-specific and pediatric-relevant classification comprises 87 records. In child mode, 55 adult-predominant records are hidden, leaving 658 records available for search. Additional knowledge-base structures include: 124 synonym entries supporting alternative symptom terminology. 29 emergency or crisis phrases used in safety-oriented detection. 104 condition records containing emergency-prefixed notes. 79 travel- or exposure-flagged records. 222 condition records with explicit NHOS Health Encyclopedia entry identifiers, corresponding to 217 distinct Encyclopedia entries. These structures support symptom retrieval, terminology matching, age-aware presentation, and contextual safety messaging. 2. Architecture and Browser Execution The widget is distributed as a standalone JavaScript component and it is integrated into NHOS web pages through a script-based deployment mechanism. Its interface is isolated using the browser's Shadow DOM, helping prevent unintended interference between the widget's presentation and the surrounding website's styles and components. The implementation supports: Standalone JavaScript deployment. Shadow DOM interface isolation. Local browser-side symptom processing. Deterministic retrieval and ranking. Direct symptom-prefill integration. Pediatric-aware presentation and hand-off. Site-wide integration across the NHOS platform. The documented standalone bundle is approximately 137.2 KB, with a compressed size of approximately 39.1 KB. The JavaScript code is approximately 21,787 bytes, while the associated knowledge-base file is approximately 115,299 bytes. 3. Symptom Retrieval and Matching Methodology The NIME™ Symptom Checker uses deterministic, rule-based symptom matching to identify condition records associated with user-entered symptoms. The retrieval process uses the structured symptom vocabulary and synonym entries to identify relevant matches. Results are ranked using the implemented matching logic and presented with the symptoms that matched each condition record. The system is designed to make its retrieval behaviour explainable: users can see which entered symptoms correspond to a returned record rather than being presented with an opaque diagnostic score. The ranking represents relative matching strength within the retrieval system. It is not a statistical probability, a measure of diagnostic certainty, or a prediction of the likelihood that a user has a particular condition. No trained statistical model is used, and no diagnostic probability is calculated or implied. 4. Safety Detection and Suggested Next Steps The widget incorporates safety-oriented phrase detection and condition-specific guidance to help users recognize situations that may warrant prompt or emergency medical attention. Its documented implementation includes: 29 emergency or crisis phrases. Emergency-prefixed notes in 104 condition records. A five-level suggested-next-step framework, ranging from self-care-oriented guidance to emergency care. Conservative heuristic rules for determining the suggested next step. Age-aware behaviour and pediatric hand-off functionality. The suggested next step is generated through predefined logic rather than a clinically validated triage algorithm. It is intended as informational guidance and must not be interpreted as a definitive assessment of medical urgency. Emergency detection and safety prompts are not exhaustive. The absence of an emergency warning does not establish that a situation is safe, and users experiencing severe or rapidly worsening symptoms should seek appropriate medical attention. 5. Pediatric-Aware Functionality The widget incorporates a child mode that changes the searchable condition set by hiding the 55 adult-predominant records. The resulting child-mode knowledge base contains 658 searchable condition records, including the pediatric-specific and pediatric-relevant records classified in the documented snapshot. The widget also provides a hand-off pathway to the NHOS Children's Symptoms Checker. The receiving checker retains its own safety-question workflow, with parents required to answer the safety questions before proceeding. This separation allows the general NIME™ widget and the dedicated children's checker to serve complementary roles within the NHOS platform. The pediatric classifications and associated content have not undergone independent clinical validation. They should therefore be interpreted as implementation-level classifications rather than clinically verified pediatric eligibility determinations. 6. Integration with the NHOS Health Encyclopedia The NIME™ Symptom Checker is integrated with the NHOS Health Encyclopedia, a structured condition-oriented knowledge resource containing 495 condition entries. The integration provides a pathway from selected widget condition records to corresponding Encyclopedia entries. The documented implementation includes: Explicit Encyclopedia entry identifiers in 222 widget condition records. References to 217 distinct Encyclopedia entries. Condition-name and alternative-name matching, supplemented by manual review. Direct entry links using the Encyclopedia's entry-identifier URL structure. A programmatic interface through which the Encyclopedia can open NIME™ with prefilled symptoms. A “Check in NIME™” interaction that transfers relevant symptoms to the widget. The integration uses explicit mapping rather than an unrestricted fuzzy-matching fallback. The two components maintain separate knowledge-base vocabularies. The widget contains 1,036 distinct symptom strings, while the Encyclopedia contains 754 distinct indexed symptom strings. Their documented exact lowercased overlap is 421 strings. This reflects differences in their respective knowledge organization and retrieval structures; exact string overlap does not capture all possible semantic equivalences. The integration is intended to connect symptom-oriented exploration with condition-oriented reference information without treating the two knowledge bases as interchangeable. 7. Privacy and Data Handling The widget is designed to process entered symptoms locally within the user's browser. Its documented implementation does not use cookies, localStorage, sessionStorage, or IndexedDB for symptom processing or storage. User-entered symptoms are not transmitted to a remote processing service by the widget. The implementation does not depend on third-party analytics services or external processing APIs for its core symptom retrieval functionality. This local-first architecture is consistent with the NHOS platform's broader privacy-oriented design principles, emphasizing user-controlled interaction and minimizing unnecessary transmission of health-related input. The privacy characteristics described here apply to the documented widget implementation. They should not be interpreted as a blanket statement about every other website component, browser extension, receiving application, or external service that a user may independently access. 8. Performance and Engineering Verification Engineering verification was performed using Playwright and Chromium to assess the documented widget's functional behaviour. The performance test measured the time required to rank 713 condition records and render six results. The reported observations were: Median execution time: 7.2 milliseconds. 95th-percentile execution time: 12.9 milliseconds. Test sample: 30 runs. Environment: headless Chromium. Measurement scope: ranking and associated DOM update. These measurements are engineering observations under the documented test conditions. They are not clinical performance measurements, diagnostic-accuracy results, or guarantees of performance across all devices and browsers. The verification process examined implementation behaviour and functionality rather than clinical correctness. 9. De