Commitment

Responsible AI

InsytAI follows a governance-first approach to AI development. Every system we build is designed with safety, accountability, and clinical trust as non-negotiable foundations — not afterthoughts.

Conflicts of Interest & Transparency

InsytAI's founder holds a concurrent senior role at a major Singapore public health institution. InsytAI advisory engagements are conducted independently and do not leverage non-public information from that employment.

Projects described in InsytAI's portfolio reflect the founder's direct technical leadership role. Institutional ownership of data, systems, and deployment authority remains with the respective healthcare organisations. InsytAI does not claim commercial ownership of systems built within institutional employment contexts.

Where the founder's institutional role and an InsytAI engagement may create a conflict of interest, InsytAI will disclose this to prospective clients and recuse from advisory activities where independence cannot be maintained.

Clinical Safety First

No system reaches production without clinical validation. We require human-in-the-loop review for any AI output that influences clinical decisions.

Explainability

Where clinically appropriate, we build explainability into models. Clinicians must be able to understand why a system reached a conclusion.

Data Protection

Patient data is treated under strict governance. We adhere to Singapore's PDPA, institutional data policies, and applicable healthcare data regulations.

Bias Monitoring

We evaluate models for performance disparities across patient subgroups — age, ethnicity, sex, comorbidity — and document known limitations explicitly.

Audit Trails

Every inference in a clinical setting must be logged with model version, input, output, and clinician action. This enables accountability and model improvement.

Continuous Monitoring

Deployed models are monitored for data drift, performance degradation, and unexpected behaviour. We do not set-and-forget clinical AI systems.

Built-In Deployment Checklist

Every InsytAI engagement follows this production readiness framework:

✦Data governance and access control documented

✦Clinical workflow mapping completed

✦Model validation against real-world data

✦Audit logs and full traceability

✦Human-in-the-loop review mechanism

✦Cloud or on-prem deployment options

✦Ethics review documentation

✦Compliance review for legal and regulatory requirements

✦Performance monitoring post-deployment

✦Clear escalation path for model failures

✦Leadership and stakeholder briefing materials

✦Clinician training and override documentation

Our Position on LLMs in Healthcare

Large Language Models offer significant potential in healthcare — but also significant risk if deployed without rigorous governance. Our approach:

Shadow mode first: LLM outputs run in shadow mode before going live, compared against clinician ground truth for a defined validation period
Hallucination monitoring: All clinical LLM outputs are monitored for factual inconsistency. We do not ship without active detection
Clinician override is frictionless: If overriding an AI suggestion takes more steps than ignoring it, the system will not be used safely. We design override into every workflow
Tiered deployment by risk: Administrative tasks deploy before diagnostic support. Risk stratification drives rollout sequence

Regulatory Classification — Software as a Medical Device (SaMD)

Clinical AI systems that influence clinical diagnosis, treatment decisions, or patient management may be classified as Software as a Medical Device (SaMD) under Singapore's Health Products (Medical Devices) Regulations, administered by the Health Sciences Authority (HSA).

InsytAI assesses each engagement against the IMDRF SaMD risk framework and advises clients on appropriate HSA regulatory pathways. Administrative AI tools (scheduling, documentation formatting) are typically exempt; diagnostic decision support tools require classification assessment before deployment.

Regulatory & Governance Frameworks

Our governance approach aligns with Singapore's regulatory and policy frameworks including:

MOH Artificial Intelligence in Healthcare Advisory (AIG framework)
IMDA/PDPC Model AI Governance Framework (Singapore)
HSA Medical Device regulations for SaMD classification
Singapore Personal Data Protection Act 2012 (PDPA, amended 2021)
IMDRF Software as a Medical Device risk classification framework

Disclaimer on Project Portfolio

InsytAI does not claim ownership of institutional AI systems unless explicitly stated. Our portfolio reflects founder-led, contributor, or research leadership roles across academic, clinical, and institutional AI initiatives. Project ownership, intellectual property, and deployment authority remain with the respective institutions.

AI-Assisted Content

Some blog posts on this website are drafted with AI assistance (Claude via AWS Bedrock) based on real published sources. AI-assisted posts are marked with an "AI-assisted" label and reviewed editorially before publication. InsytAI does not publish AI-generated clinical claims without editorial review.

All research sources cited in AI-assisted posts are drawn from peer-reviewed literature (PubMed, Europe PMC, Semantic Scholar), preprint servers (arXiv, medRxiv, bioRxiv), major journals (Nature, The Lancet, JAMA, NEJM), editorial platforms (Medium — clearly labelled), and verified news RSS feeds. Citations are verifiable via the linked sources.

Contact

Questions about our responsible AI practices: info@insytai.com