# InsytAI

> Clinical AI systems, healthcare analytics, and AI governance, built for real health-system deployment.
> We help healthcare teams design, evaluate, and deploy clinical AI workflows, analytics platforms, RAG systems, and AI governance frameworks.

## About
- Website: https://new.insytai.com/
- Blog: https://new.insytai.com/blog.html
- Work: https://new.insytai.com/work/
- Full content index: https://new.insytai.com/llms-full.txt
- Focus: clinical AI & prediction models, healthcare data platforms, RAG/LLM workflows, AI governance & evaluation
- Geography: Singapore, Southeast Asia, APAC

## Key pages
- Home: https://new.insytai.com/
- Blog index: https://new.insytai.com/blog.html
- Responsible AI: https://new.insytai.com/responsible-ai.html
- Privacy: https://new.insytai.com/privacy.html

## Recent articles
- [LLM Governance in Healthcare: Building Responsible AI Frameworks for Clinical Deployment](https://new.insytai.com/blog/llm-governance-healthcare-2025): Lessons from authoring a hospital cluster-wide LLM Strategy Paper — what governance frameworks must cover before any language model touches clinical data.
- [Multi-Agent AI Systems in Clinical Workflows: From Research to Reality](https://new.insytai.com/blog/multiagent-systems-clinical-workflows): How autonomous agent networks can transform hospital operations — and the hard lessons from deploying them in production at major Singapore health clusters.
- [Deep Learning for Medical Imaging in Southeast Asia: Dataset Challenges and Solutions](https://new.insytai.com/blog/medical-imaging-deep-learning-singapore): Training medical imaging models on Asian patient populations requires different approaches. Here's what we've learned across 8 imaging AI projects at Singapore's major hospital clusters.
- [The Patient Voice in Healthcare AI: Why Asia's Hospitals Need Structured Feedback Mechanisms](https://new.insytai.com/blog/the-patient-voice-in-healthcare-ai-why-asias-hospitals-need-structured-feedback-): Stanford's patient panels reveal critical gaps in AI adoption. Asia's healthcare systems must build similar feedback loops before deployment.
- [The Invisible Layer: Why Medical Imaging AI Fails Before You See the Output](https://new.insytai.com/blog/medical-imaging-ai-acquisition-drift-monitoring-singapore): Governance frameworks monitor AI outputs. But a new study shows lung-nodule AI fails at the acquisition layer—invisible to DICOM metadata and current monitoring.
- [Singapore's AI-SaMD Exemption Pathway: What Public Healthcare Institutions Need to Know in 2026](https://new.insytai.com/blog/singapore-ai-samd-exemption-pathway-hsa-public-healthcare-2026): HSA's new exemption pathway for AI-SaMD in public healthcare institutions changes the regulatory game—but only if you understand the constraints.

## Portfolio case studies
- [RUSSELL GPT](https://new.insytai.com/work/russell-gpt.html): Fine-tuned clinical LLM for discharge summary generation. Published as a real-world RUSSELL GPT pilot study in The Lancet Regional Health–Western Pacific 51 (20
- [CalSense+](https://new.insytai.com/work/calsense.html): AI detection of calcium abnormalities for hyperparathyroidism diagnosis and treatment planning. $3.97M MOH-funded. CO-PI.
- [MediVoice](https://new.insytai.com/work/medivoice.html): Medical voice transcription and clinical data reformatting platform. Converts unstructured voice notes into structured clinical records.
- [MRI Spine Auto-Protocoling LLM](https://new.insytai.com/work/mri-spine-llm.html): Secure institutional LLM for MRI spine request form enhancement and auto-protocoling. Published in The Spine Journal 25(3), 505-514 (2025).
- [SMILE AI](https://new.insytai.com/work/smile-ai.html): Smart Modelling and Intelligent Learning for Enhancing Oral Health. Automated oral health diagnosis platform for population-level screening.
- [HoloVein](https://new.insytai.com/work/holovein.html): Published mixed-reality venipuncture aid using CNNs and semi-supervised learning, with HoloLens vein overlay for guided IV access.
- [HoloPOCUS](https://new.insytai.com/work/holopocus.html): Portable mixed-reality 3D ultrasound tracking, reconstruction and overlay system. Published at ASMUS/MICCAI 2023 and in Lecture Notes in Computer Science.
- [Discovery AI Platform](https://new.insytai.com/work/discovery-ai.html): System architect for production-grade clinical AI deployment platform integrating with hospital cluster EMR, data governance SOPs, and multi-department data fee
- [Scoliosis AI](https://new.insytai.com/work/scoliosis-ai.html): Published automated Cobb angle measurement for scoliosis radiographs, with GAN-based vertebrae segmentation and deep learning-assisted productivity work.
- [Dynamic Scheduling for Healthcare](https://new.insytai.com/work/dynamic-scheduling.html): Machine learning models to optimize patient scheduling in clinics and other healthcare settings.
- [Allergic Comorbidities — Singapore & Australia Cohorts](https://new.insytai.com/work/allergic-comorbidities.html): Cohort study examining allergic disease risk across ancestry and early-life environmental contexts in Singapore and Australia.
- [Cognitive Function Prediction in Preterm Children](https://new.insytai.com/work/cognitive-function-prediction.html): Neuroimaging and machine-learning study using brain connectivity features to support cognitive-risk stratification in preterm children.
- [Pancreatitis Severity Prediction](https://new.insytai.com/work/pancreatitis-severity.html): Clinical prediction work evaluating severity scoring for acute pancreatitis and supporting risk stratification in hepatobiliary care.
- [Blood Culture Outcome Score — Oncology](https://new.insytai.com/work/blood-culture-score.html): Prediction model to estimate risk of positive blood cultures during chemotherapy-related oncology care.
- [Functional High-Risk Myeloma Prediction](https://new.insytai.com/work/functional-high-risk-myeloma.html): Genomic model to identify functionally high-risk multiple myeloma patients using transcriptomic and clinical features.
- [Unified Breast Cancer Cohort — Singapore](https://new.insytai.com/work/breast-cancer-cohort-sg.html): Unified breast cancer research database with NLP tools to extract structured data from pathology reports, imaging notes, and clinical documentation.
- [Breast Cancer Cohort Enrichment Platform](https://new.insytai.com/work/cancer-cohort-enrichment.html): Machine-learning workflow to enrich breast-cancer screening cohorts for early-detection research and prioritisation.
- [Endeavour AI (EAI Platform)](https://new.insytai.com/work/endeavour-discovery-ai.html): Production platform integrating next-generation EMR feeds with governed clinical AI applications via secure interoperability and containerised deployment.
- [PRESCIENCE HPC Platform](https://new.insytai.com/work/prescience.html): On-premises high-performance computing platform enabling secure deep-learning workloads for sensitive healthcare data within institutional boundaries.

## Contact
- Enquiries: https://new.insytai.com/#contact