The next major advance in healthcare AI may not be another diagnostic chatbot or clinical documentation tool. It could be an AI system capable of recognizing subtle changes in a person's health years before conventional medicine would typically intervene.

That is the ambition behind a new research partnership between Function and NYU Grossman School of Medicine, part of NYU Langone Health. The organizations plan to combine longitudinal health measurements with clinical data to investigate whether AI can identify early signals associated with disease and help clinicians intervene sooner.

The partnership brings together two different views of health. NYU Grossman has extensive longitudinal clinical information showing how diseases develop, are diagnosed and treated among patients who have required medical care. Function, meanwhile, collects advanced laboratory testing and imaging data intended to provide a broader view of health across people both before and after disease emerges.

That distinction could be important for developing predictive healthcare AI. Medical datasets often contain extensive information about patients once they become sick, but identifying disease before diagnosis requires information about what happens during the less visible period leading up to it.

Function says the collaboration will use these complementary datasets to develop and evaluate AI models focused on earlier disease detection and clinical decision support.

The initial research agenda includes three areas: early cancer warning, brain-health monitoring and cardiometabolic risk.

For cancer, researchers will look for subtle signals that could potentially identify disease at an earlier stage, when treatment may be more effective. The brain-health work will examine changes over time that could provide insights into cognitive risk. Cardiometabolic research will explore methods for assessing future heart and metabolic risks earlier and supporting more personalized interventions.

The underlying technology is longitudinal multimodal health analysis. Instead of evaluating one blood test, scan or clinical encounter in isolation, machine-learning systems can potentially analyze patterns across multiple measurements collected over time.

That approach is particularly relevant to preventive medicine. Human clinicians can identify many risk factors, but tracking thousands of variables across years is difficult. AI can potentially detect relationships between measurements that are difficult to identify manually, provided the underlying data is sufficiently broad, reliable and clinically validated.

Daniel K. Sodickson, chief medical scientist at Function and former chief of innovation in radiology at NYU Langone Health, is helping lead the initiative. He has framed the partnership around using advanced testing earlier rather than primarily after illness has already emerged.

Michael P. Recht, Louis Marx Chair of Radiology at NYU Grossman School of Medicine, described the collaboration as an opportunity to combine academic clinical research with newer approaches to longitudinal health measurement.

The partnership also expands Function's Medical Intelligence Lab, introduced in November 2025. The lab is focused on applying machine learning and medical expertise to longitudinal health information, with the broader goal of helping people understand and act on potential health risks earlier.

The commercial healthcare technology market already includes major AI players working on medical imaging, diagnostics, clinical decision support and drug discovery. Companies such as Microsoft, Google and NVIDIA are also investing heavily in healthcare AI infrastructure, while specialized firms are developing models for radiology, pathology and clinical workflows.

The challenge for Function and NYU Grossman is therefore not simply building another healthcare AI model. It is proving that longitudinal predictive models can deliver clinically meaningful signals without generating excessive false positives or creating unnecessary medical interventions.

That distinction is critical. A system designed to warn people about potential disease must balance sensitivity with specificity. Detecting every possible anomaly may sound useful, but excessive alerts can lead to unnecessary testing, anxiety and increased healthcare costs.

Scientific validation will consequently be central to the partnership. Function says tools, models and peer-reviewed publications produced through the research will be subjected to rigorous scientific standards.

The data strategy will also matter. Healthcare AI models are highly dependent on the diversity and quality of the datasets used for development and validation. A model trained primarily on one population or clinical environment may not perform equally well across different demographic and health groups.

This is where NYU Grossman's academic role could become particularly important. Clinical research can provide the validation frameworks needed to determine whether patterns discovered in large datasets actually correspond to meaningful disease risks.

The partnership arrives during a period of expansion for Function. The company recently acquired Getlabs and SuppCo and secured $450 million in growth financing from General Catalyst's Customer Value Fund. Those moves give Function additional resources as it attempts to build a broader health-monitoring platform.

The bigger industry implication is a potential shift in the role of AI in medicine. Much of today's healthcare AI focuses on making existing workflows faster, such as summarizing clinical notes, interpreting images or assisting administrative tasks. Predictive health AI represents a different ambition: using accumulated data to identify risk before conventional clinical pathways trigger intervention.

If researchers can demonstrate that these models reliably identify clinically useful signals early, the implications could extend across preventive care, screening, personalized medicine and chronic disease management.

For now, the Function-NYU Grossman collaboration remains a research initiative rather than evidence that AI can predict individual diseases with clinical certainty. Its significance lies in the attempt to build and rigorously evaluate the data infrastructure and models required to make earlier detection scientifically credible.

The most important milestone will not be the launch of an AI model. It will be evidence that the model can improve real clinical decisions.

Market Landscape

Healthcare AI is moving from narrow workflow automation toward increasingly ambitious applications in diagnostics, risk prediction and personalized medicine.

Companies across the technology and healthcare sectors are investing in AI for medical imaging, clinical decision support, drug discovery and patient engagement. Microsoft, Google and NVIDIA are among the major technology companies building healthcare AI infrastructure, while specialized medical-AI companies focus on specific clinical applications.

Function and NYU Grossman's approach differs in its emphasis on longitudinal multimodal data. Rather than focusing on a single clinical event, the research aims to understand how combinations of laboratory, imaging and health measurements change over time.

That model could support a more preventive approach to healthcare, but it also raises substantial challenges around data quality, validation, privacy, bias and false positives.

Strategic Outlook

The partnership points toward a healthcare model in which AI increasingly acts as a longitudinal risk-monitoring layer rather than simply a tool used during a clinical encounter.

Function's Medical Intelligence Lab provides the company's technology and research framework, while NYU Grossman brings academic medicine and clinical research expertise.

The opportunity is significant, but so is the burden of proof. Predicting disease earlier requires models that are accurate across diverse populations and capable of demonstrating that earlier warnings actually improve outcomes.

If that evidence emerges, longitudinal AI could become an important component of preventive health infrastructure.

Top Insights

• Function and NYU Grossman are combining longitudinal health data and clinical expertise to investigate AI models for earlier disease detection.
• The research will initially examine cancer signals, brain-health changes and cardiometabolic risk, targeting conditions where earlier intervention could influence outcomes.
• Longitudinal multimodal AI could identify health patterns across laboratory tests and imaging that conventional single-visit assessments may miss.
• Scientific validation will be critical because predictive healthcare AI must balance early warnings against false positives and unnecessary medical interventions.
• The partnership reflects healthcare AI's broader evolution from workflow automation toward preventive risk prediction and longitudinal health intelligence.

Get in touch with our MarTech Experts