Research & Datasets

Data and Clinical-Rule Infrastructure for Healthcare AI in Asia

We believe reliable healthcare AI rests on deterministic clinical rules and real-world data. HealthyMind Tech is building a longitudinal research database focused on dietary behavior and clinical outcomes, and open-sources Taiwan-Health-MCP medical-standard tools along with a deterministic clinical-logic evaluation environment.

500K
Target food images
10,000
Target patients tracked longitudinally
60+ days
Continuous tracking span
FHIR R4
International health data standard
Core Datasets

Three Core Dataset Programs

From image recognition to clinical outcomes, three dataset layers form a closed loop for research on Asian diets and chronic disease.

Layer 01
Program launching

Taiwanese Food Image Recognition Set

Global healthcare AI struggles to recognize distinctly Asian meals such as Taiwanese bento boxes and braised pork rice. This set builds a high-accuracy foundational recognition model.

  • Target size: 500,000 photos
  • Annotation: Calories and six major nutrients labeled by dietitians
  • Taxonomy: Taiwan Food Composition Database
Layer 02
IRB application in planning

Longitudinal Dietary Behavior Database

Using the high retention of the DietMate LINE platform, we capture months of real-world eating patterns, avoiding the bias of traditional recall methods.

  • Target size: Tracking data from 10,000 patients
  • Time span: 60+ consecutive days
  • Analysis focus: Meal timing, nutrient distribution, adherence curves
Layer 03
Hospital partnerships in preparation

Diet × Chronic Disease Outcomes RWE Database

Combines out-of-hospital behavioral data with in-hospital lab results to provide comprehensive real-world evidence (RWE) for drug development and PSP evaluation.

  • Clinical linkage: HbA1c, eGFR, systolic blood pressure
  • Technical standard: FHIR R4 (Observation, Condition)
  • Privacy: Medical-grade de-identification pipeline
Architecture & Standards

Medical-Grade Rigor and Standards

DietMate LINE BotPatient meal logs · out-of-hospital behavioral data
Taiwan Health MCP & CQL51+ medical-standard tools · deterministic clinical rules
De-ID PipelineHIPAA Safe Harbor-guided de-identification
FHIR R4 LayerObservation · Condition · Patient
Clinical Data LinkageHbA1c · eGFR · systolic BP · hospital EMR

A strong database starts with the right foundation. We adopted international standards from day one so every record can feed rigorous medical research and drug evidence.

  • FHIR R4

    Built on the international standard for health data exchange, ensuring interoperability and cross-system integration.

  • IRB Review

    Every retrospective and prospective study requires approval from an institutional review board. Research ethics are non-negotiable.

  • De-ID

    A de-identification pipeline designed around the HIPAA Safe Harbor method balances research value with patient privacy.

  • RWE

    Real-world data collection and analysis are designed with reference to the FDA RWE framework, aiming to provide real-world evidence for clinical research.

Live Query Simulation

Live Clinical Dataset Query

Query the longitudinal status of diet, nutrients, and clinical indicators through standard FastMCP and FHIR R4 APIs.

healthymind-cohort-query
❯

Explore a Partnership

We are actively seeking partnerships with academic institutions, clinical trial centers, and global pharmaceutical companies to unlock the potential of real-world evidence (RWE).

Contact the Research Team Back to Home