III. Product & Technology Adaptation Strategy
Jianpei Technology's core advantage lies in its AI image-reading technology, but the key to success in Southeast Asia is the deep integration of technology with localized needs, following the path of "Technology Adaptation + Local Co-creation".
3.1 Localization & Co-creation Mechanism
Jianpei Technology needs to start with large-scale Proof of Concept (PoC) to test and adapt AI solutions within local clinical scenarios. This requires demand discussions and field research with local institutions and experts to ensure that algorithms and features are adjusted around actual clinical pain points.
Case Study: Indonesia
Market Insight
In Indonesia, cardiovascular disease is one of the greatest health challenges, with cases reaching 20.04 million in 2023.
Execution Strategy:
- Partner with local cardiovascular centers (e.g., hospitals where Siemens has established partnerships).
- Utilize local datasets to train and validate AI models for high-prevalence diseases in the region.
- Ensure AI models can accurately identify specific pathological features and imaging equipment variances of the local population.
Co-creation Value
Localization is not just about translating the software interface; it is about ensuring the AI model accurate identifies regional pathological nuances. This approach transforms partners from pure "technology receivers" to "solution co-builders".
3.2 Core Product System: "Dr. Wood" AI-CDSS Deployment in SEA
"Dr. Wood" is Jianpei Technology's core product, evolving from the first image-reading robot to the first "practicing" AI doctor, covering the full Prevention-Screening-Diagnosis-Treatment-Rehabilitation cycle.
3.2.1 Product Architecture Layers
| Layer | Core Content | SEA Adaptation Points |
|---|---|---|
| Data Layer | Medical knowledge (diseases/drugs/treatments), clinical data (HIS/LIS/PACS), health data | Interface with local HIS systems, establish regional data lakes |
| Tech Layer | Dr. Wood Medical LLM (500B+ tokens), Medical Knowledge Graph, Deep Learning Engine | Multi-language support (English, Malay, Indonesian), private deployment |
| Product Layer | Single-disease AI-CDSS full-lifecycle intelligent medical services | Prioritize deployment based on regional high-prevalence diseases |
3.2.2 Full-process Solution (Phased Deployment)
| Phase | Functional Module | Deployment Priority |
|---|---|---|
| Pre-diagnosis (P, S) | Risk assessment, intelligent triage, health alerts, early cancer screening | ⭐⭐⭐ High (Primary care entry point) |
| Intra-diagnosis (D, T) | AI-assisted diagnosis, EMR generation, medication advice, surgical navigation | ⭐⭐⭐ High (Core value) |
| Post-diagnosis (R) | Follow-up management, behavioral intervention, patient management, health management | ⭐⭐ Medium (Value-added service) |
3.2.3 Diagnosis Process Quality Control (Differentiated Competitiveness)
| Step | QC System | Functional Description |
|---|---|---|
| Capture | Image QC | Intelligent analysis of exposure, artifacts, contrast, resolution, SNR |
| Diagnosis | Diagnostic QC | Evidence-based diagnostic suggestions, potential error alerts, standardized clinical paths |
| Report | AI Structured Reporting | Automatic extraction of key info, mandatory standardized fields, terminology validation |
| EMR | EMR QC | NLP parsing of unstructured text, logical consistency checks, real-time defect alerts |
3.2.4 Specialty-specific Treatment & Intelligent Device Ecosystem
| Specialty Area | Core Functionality | SEA Use Case |
|---|---|---|
| Digital Lung | Intelligent lung nodule diagnosis (99% detection for 3mm+), precise lesion quantification | TB high-prevalence areas (Indonesia, Philippines) |
| Digital Heart | Coronary CTA image analysis, arterial plaque/stenosis detection | Cardiovascular disease prevalence (Malaysia) |
| Digital Liver | Liver elasticity imaging, fatty liver/fibrosis detection | Hepatitis high-prevalence regions |
| Smart Wearables | Data integration from BP monitors, glucometers, ECG, portable ultrasound, etc. | Remote monitoring, primary healthcare |
Wuzhen Model Replicability
The Wuzhen Intelligent Hospital has validated the "Central Hub + N Intelligent Clinics" model. Launched in June 2024, it already has 50+ affiliated clinics and 20,000+ visitors. This model is highly replicable in Sabah, Malaysia.
3.3 Data Governance & AI Architecture Requirements
Given the sensitivity of medical AI, the technical architecture must meet the most stringent regulatory requirements in the region.
HSA GL7 Core Focus Areas
Singapore's HSA GL7 guidelines specifically focus on:
| Focus Area | Requirement |
|---|---|
| Continuous Learning | Control of model adaptation capabilities |
| Automation Level | Degree and timing of human intervention |
| Model Retraining | Update protocols and validation procedures |
| User Intervention | Final decision-making power of clinicians |
Technical Preparation
::: important Compliance Requirement Technical preparation must include: establishing continuous performance monitoring and real-world evidence collection mechanisms, with regular reporting to regulators like the HSA. :::
Key Architectural Decisions:
- AI model updates and maintenance cannot be simply deployed from China.
- Regionalized data pipelines and auditing capabilities are required.
- Ensure data security and compliance in model iterations.
AI Verify Framework
Utilizing Singapore's AI governance testing framework and software toolkit, AI Verify (published by IMDA and PDPC), can systematically assess:
- Transparency
- Ethics
- Security
Change History
| Version | Date | Description | Author |
|---|---|---|---|
| 1.0 | 2025-12-16 | Initial Version | Gemini |
| 1.1 | 2025-12-16 | Updated "Dr. Wood" AI-CDSS product system; added full-process QC, specialty-specific treatment, and Wuzhen model. | Gemini |