Medical AI model audit
Independent technical review of SaMD, SaaS clinical AI, diagnostic/prognostic models, LLM/RAG systems, and on-device medical AI.
Consultancy & Advisory
I help biomedical, pharma, MedTech, digital health, and clinical research teams build, validate, audit, and repair AI/ML pipelines for diagnostic/prognostic models, biomarkers, MLE-NITs, SaMD, SaaS clinical AI, LLM-based decision support, and on-device medical AI.
Primary commercial focus
The most focused consultancy offer is independent AI/ML model audit and regulatory-facing evidence support: leakage, confounding, validation design, calibration, subgroup behavior, reproducibility, XAI, deployment assumptions, and lifecycle risks.
Independent technical review of SaMD, SaaS clinical AI, diagnostic/prognostic models, LLM/RAG systems, and on-device medical AI.
Leakage, cohort effects, batch effects, confounding, overfitting, split design, thresholding, and reproducibility checks.
Scientific and technical support around validation, transparency, model documentation, intended-use alignment, and lifecycle risk thinking.
Independent review of medical AI startup claims, validation evidence, scientific credibility, deployment assumptions, and regulatory-facing gaps before investment decisions.
Models for diagnosis, prognosis, biomarkers, imaging, omics, tabular clinical data, and clinical decision support.
Machine-learning-enhanced non-invasive tests: compact, interpretable upgrades to clinical scores and biomarker panels.
Biomarker prioritization, biological signal retrieval, and potential drug-target identification for pharma-relevant discovery.
ZACH-ViT, hZACH-ViT, s-DNNs, and compact baselines that can be trained and audited without massive compute.
RAG, clinical decision-support prototypes, drug-safety workflows, and multimodal pipelines combining text, imaging, omics, and structured data.
Angel investors, venture capital, family offices, corporate innovation
I can support investment teams evaluating medical AI, SaMD, digital health, biomarker, clinical decision-support, and on-device AI companies. The goal is to independently assess whether the company's AI claims are technically credible, clinically meaningful, reproducible, bias-aware, and supported by evidence that matches the intended use.
Typical checks include data provenance, validation design, patient/site leakage, external validation, calibration, subgroup performance, explainability, deployment assumptions, regulatory-facing evidence gaps, and the realism of the startup's claimed performance.
I combine mathematical/statistical methodology, biomedical AI research, high-stakes model validation, explainable AI, and production data science experience across Amsterdam UMC, Diagnostic Echotomography, MGH/Harvard, Huawei, Plotwise, and independent biomedical AI consultancy.
Useful when a project needs rigorous validation, reproducible code, clear interpretation, and defensible technical documentation.
Scope note
I provide scientific, statistical, and machine-learning evidence support for regulatory-facing, clinical, investor, and production-readiness decisions. This does not replace notified bodies, legal counsel, quality-management specialists, or regulatory affairs consultants.
Use this form for model audit, consultancy, seminars, or collaboration
Briefly describe your organization, model/system, data modality, intended use, current validation stage, and timeline.
Available globally in English and Greek - online, onsite, or hybrid.
Contact
For confidential or early-stage projects, send a short non-confidential summary first.