Skip to content
#small language model Open access

NHOS Children's Symptoms Checker™ A Client-Side Naive Bayes Pediatric Triage Aid with a 50-Level Triage Severity Score, 3,392-Condition Library, 17-System Coverage, Integrated Interactive Visualization, and a Nine-Module Complete Clinical Upgrade Pack Technical Research White Paper — Architecture, Methodology, Validation Framework, Reproducibility Specification, and Complete Clinical Upgrade Manifest

Sep 2026 · Zenodo (CERN European Organization for Nuclear Research)

Abstract

The NHOS Children's Symptoms Checker™ (NHOS-CSC) is a fully client-side pediatric symptom-triage artifact released as informational research software. It is designed to answer a narrow methodological research question: can a compact, fully inspectable, browser-local pediatric symptom-triage tool expose its assumptions and computational pathway sufficiently for independent audit, replication, and critique — while still delivering a graded, safety-first triage recommendation? This release (v4.0.0) documents the artifact, its nine-module clinical upgrade pack, and its proposed validation framework. What the artifact is NHOS-CSC is architected for fully client-side execution: all inference runs locally within the client runtime, with no server-side inference, no hosted API dependency, and no embedded analytics. Its knowledge base — 3,392 unique pediatric conditions and 591 clinically weighted symptom attributes across 17 body systems — is parsed and indexed at application startup. Its inference engine implements a Naive Bayes classifier in log-space with age-aware condition filtering, explicit negative-evidence handling, a small co-occurrence adjustment, and posterior normalization. Output is a discrete 50-level Triage Severity Score (0–49) mapped to six clinically meaningful care bands: Self-Care, Routine, Soon, Urgent, Emergency, and Emergency-Ambulance. A mandatory 19-question red-flag screen intercepts life-threatening presentations before any statistical inference occurs. What is new in v4.0.0 — the Complete Clinical Upgrade Pack Version 4.0.0 introduces a nine-module Complete Clinical Upgrade Pack composed through an additive composition layer that wraps four well-defined core extension points without modifying the underlying inference, scoring, or knowledge-base modules. Zero lines of the core application are modified. The upgrade is fully reversible by removing a single add-on layer and is fully configurable through a single public configuration object exposing 13 documented tuning parameters. The nine modules are: Pediatric Vitals Interpretation Engine — age-banded physiological reference ranges across six pediatric cohorts (neonate, infant, toddler, preschool, school-age, adolescent) for heart rate, respiratory rate, systolic blood pressure, temperature, and peripheral oxygen saturation. Critical vitals readings escalate the triage score through hard constraints. Expanded Red-Flag Detection — the mandatory pre-inference safety screen expands from 9 to 19 binary questions, adding bulging fontanelle, petechiae, stridor, lethargy, poor neonatal feeding, toxic appearance, bilious vomiting, bloody stool, active seizure, and purpura. Developmental Milestone Reference — 16 age-matched milestone bands from 1 month to 17 years, displayed within a configurable ±2-year window around the child's age. Medication Safety Layer — 8 age-filtered medication entries with weight-aware dosing context and hard-coded aspirin contraindication for children under 16. Parental Guidance UX Layer — plain-language actionable guidance translating triage output into immediate steps and emergency escalation criteria. Clinician Dashboard Modal — a single-view clinical summary combining organ-system heatmap, ranked differential (top 6), vitals flags, predictive indices, and case timeline. Four Predictive Risk Indices — heuristic 0–10 screening scores covering early sepsis risk, respiratory distress, dehydration, and febrile illness risk. Explain-This-Result Panel — a plain-language "Why this band?" panel surfacing the top 2–5 contributing factors behind the assigned triage band. Clinical Content Versioning System — nine independent content version tracks that separate clinical knowledge versions from software release versions, enabling independent audit and staged clinical-content upgrades. Architecture and reproducibility The artifact preserves every architectural guarantee of the prior releases: client-side-only execution, zero server dependency, full inspectability, and no telemetry. All processing — condition parsing, symptom indexing, red-flag screening, vitals interpretation, Bayesian inference, triage scoring, milestone generation, medication filtering, guidance generation, predictive index computation, explanation reason extraction, and content versioning — executes locally within the client runtime. The only network requests are optional presentation resources (fonts, icons, and visualization libraries); none of them receive user inputs. A runtime library audit is emitted on application startup, reporting declared design targets versus actual loaded counts for every headline metric. The Triage Severity Score is fully specified by six base values, five adjustment terms, five hard constraint rules (including one vitals-derived), one integer clamp, six band boundaries, and eight vitals escalation thresholds — enabling bit-exact independent reproduction. What the artifact is not This artifact is not a medical device and is not cleared or approved by the FDA, EMA, MHRA, TGA, or any other regulatory authority. It has not been evaluated in a prospective clinical trial, subjected to retrospective chart review, validated against a gold-standard vignette set, or calibrated against clinical outcomes. The probabilities, rankings, and triage recommendations produced by the model are model-derived quantities, not clinically calibrated performance measures. They must not be interpreted as diagnostic, prognostic, or therapeutic guidance. In any emergency, contact emergency services immediately — 911 (US) · 999 (UK) · 112 (EU) · 000 (AU) — regardless of what this artifact displays. What this deposit contains The Zenodo record for this release contains the reference implementation of the artifact plus the Complete Clinical Upgrade Pack, the machine-readable condition and symptom libraries, the configuration schema reference, the triage-score specification, the vitals reference range specification, the milestone band specification, the medication reference specification, the predictive risk index specification, the clinician dashboard specification, the explain-this-result specification, the content versioning specification, the proposed validation protocol, the technical white paper (PDF and Markdown), a README with execution and tuning instructions, and license files (MIT for code; CC BY 4.0 for documentation). Intended use The artifact is offered as: An informational triage aid for parents and caregivers. A teaching artifact for health informatics and medical education. A reproducible research baseline for symptom-checker methodology. A transparent pediatric baseline for comparative evaluation. A worked example of client-side clinical decision support with integrated interactive visualization. It is explicitly not intended for diagnosis, replacement of clinical judgment, use as a regulated clinical workflow, or population-level epidemiological inference. Citation and licensing Software code is released under the MIT License. Documentation and research text are released under Creative Commons Attribution 4.0 International (CC BY 4.0). This artifact should be cited as software and methodology research. Its future scientific value will depend on independent review, empirical validation, calibration analysis, and transparent reporting of both successful and unsuccessful results.

View source

Similar papers

#computer vision Open access Jun 2016

Software Development in Startup Companies: The Greenfield Startup Model

The results are packaged in the Greenfield Startup Model (GSM), which explains the priority of startups to release the product as quickly as possible, and the need to shorten time-to-market, by speeding up the development through low-precision engineering activities.

Carmine Giardino, Nicolò Paternoster, M. Unterkalmsteiner et al. · 178 citations · ⚡14
#computer vision Open access Oct 2016

Software Startups - A Research Agenda

Software startup companies develop innovative, software-intensive products within limited timeframes and with few resources, searching for sustainable and scalable business models.

M. Unterkalmsteiner, P. Abrahamsson, Xiaofeng Wang et al. · 157 citations · ⚡17
#machine learning Review Open access Oct 2016

“Failures” to be celebrated: an analysis of major pivots of software startups

This study conducts a case survey study based on the secondary data of the major pivots happened in 49 software startups, and demonstrates that customer need pivot is the most common among all pivot types.

Sohaib Shahid Bajwa, Xiaofeng Wang, Anh Nguyen-Duc et al. · 127 citations · ⚡15
#computer vision Review Open access May 2015

A survey study on major technical barriers affecting the decision to adopt cloud services

The comparison of adopter and non-adopter sample reveals three potential adoption inhibitor, security, data privacy, and portability, which underlines the importance of the technical and security perspectives for research investigating the adoption of technology.

Nattakarn Phaphoom, Xiaofeng Wang, S. Samuel et al. · 111 citations · ⚡8
#computer vision Open access Feb 2018

Lean Internal Startups for Software Product Innovation in Large Companies: Enablers and Inhibitors

This study investigates how Lean internal startup facilitates software product innovation in large companies and identifies its enablers and inhibitors, and shows the potential of the method-in-action framework to investigate the Lean startup approach in non-startup context.

Henry Edison, Nina M. Smørsgård, Xiaofeng Wang et al. · 78 citations · ⚡6
#computer vision Conference Sep 2010

Exploring the Sources of Waste in Kanban Software Development Projects

The application of agile software methods and more recently the integration of Lean practices contribute to the trend of continuous improvement in the software industry. One such area warranting proper empirical evidence is a project’s operational efficiency when using the Kanban method. This short paper takes a new angle and explores waste in the Kanban-driven software development project context. A preliminary research model is presented for helping the consequent replication of the study. The results from the empirical analysis suggest Kanban can be an effective method in visualizing and organizing the current work, but does not prevent waste from creeping in, although the overall project outcome may be successful.

Marko Ikonen, Petri Kettunen, Nilay V. Oza et al. · 67 citations · ⚡9

Related blog posts

MIT News · Artificial Intelligence Sep 24, 2026

Estimating suicide risk from text

A new language-processing tool could help identify the highest-risk individuals from natural language, enabling swifter interventions.

Microsoft Research Blog Sep 21, 2026

Improving synthesis prediction of small molecules at scale with RetroChimera

Custom-made molecules are advancing medicine, materials, and agriculture, but producing them is slow and expensive. A new Nature paper highlights RetroChimera, a predictive model that helps accelerate chemical synthesis, helping researchers explore a wide range of molecules. The post Improving synthesis prediction of small molecules at scale with RetroChimera appeared first on Microsoft Research.

MIT News · Artificial Intelligence Sep 14, 2026

New method enables AI for safety-critical situations

The “HardFlow” algorithm could help generative AI models produce high-quality outputs that obey strict requirements when “pretty close” doesn’t cut it.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.