How AI can help detect disease and accelerate medical breakthroughs
AI is moving from theory to bedside across genomics, pathogen tracking and drug design. Still, separating useful tools from noise is hard when you're running a clinic, a lab or a public health program.
Rice University's AI2Health research cluster, supported by the Ken Kennedy Institute, brings together computational biology, machine learning and systems biology to build practical, interpretable tools that support real decisions in care and public health.
What AI2Health focuses on
- DNA-based modeling to forecast complex diseases like Alzheimer's and dementia.
- Pathogen surveillance for infectious disease tracking and pandemic mitigation.
- Computational oncology to improve cancer detection and targeting.
- AI-enabled vaccine and drug design to shorten R&D cycles and sharpen hit selection.
Experts you can learn from
- Todd Treangen - Biosecurity and biosurveillance: Builds machine learning algorithms and open-source software to rapidly identify harmful pathogens in synthetic DNA and metagenomic data. Work supports outbreak response and infectious disease monitoring. Lead researcher for AI2Health.
- Vicky Yao - Multi-omic methods for deciphering health and disease: Develops interpretable machine learning and statistical approaches that integrate diverse datasets to reveal molecular drivers of complex diseases such as cancer and Alzheimer's.
- Santiago Segarra - AI and machine learning for genomics and metagenomics: Advances graph machine learning to model protein interactions, genetic organization and microbial ecology in large-scale biological systems.
- Ivan Coluzza - Computational biophysics for biomedical innovation: Uses physics-based models to study protein function and molecular design, including biomimetic materials inspired by protein folding.
- Cameron Glasscock & Lydia Kavraki - Computational and systems biology for next-generation therapeutics: Combine protein design, synthetic biotechnology and AI-enhanced modeling to engineer new functions. Kavraki develops algorithms and software to model protein flexibility and function, speeding drug discovery and enabling more precise, personalized cancer immunotherapies.
- Luay Nakhleh - Evolutionary biology: Creates computational methods to study how genes, genomes and cellular networks evolve, with applications that clarify disease onset, progression and cancer genomics.
- Fritz Sedlazeck - Human genomics and structural variation: Builds next-generation AI and machine-learning methods to decode the full spectrum of human genomic variation, improving diagnoses, personalizing risk prediction and revealing mechanisms behind neurological, cardiovascular and developmental disorders.
"As a computational biologist, I think the field is at an interesting inflection point, and we can expect to see significant gains in the speed and scale at which we can analyze genomic data and uncover biological insights," said Nakhleh, the William and Stephanie Sick Dean of Rice's George R. Brown School of Engineering and Computing and professor of computer science and biosciences. "Continued collaboration and attention to the ethical dimensions of these tools will be essential going forward, and that commitment is at the core of the AI2Health research cluster."
Practical takeaways for healthcare leaders
- Strengthen pathogen surveillance: Pair wastewater or clinical sampling with AI-assisted metagenomic analysis to detect emerging threats earlier. Build vendor-agnostic pipelines and set clear QA/validation criteria. See CDC guidance on wastewater surveillance here.
- Make multi-omics actionable: Use interpretable models that integrate genomics, transcriptomics and proteomics to refine differential diagnoses and stratify risk. Prioritize tools with transparent features and audit trails.
- Upgrade oncology workflows: Combine computational pathology/radiology with molecular tumor boards to sharpen treatment selection. Tie outputs to guidelines and prospective validation before routine use.
- Accelerate therapeutics-safely: Leverage protein design and in-silico screening to narrow candidates, then confirm with wet-lab cycles. Monitor model drift and maintain versioned datasets and protocols.
- Embed biosecurity checks: Screen synthetic DNA orders and research constructs with AI tools to flag concerning sequences. Document review processes and escalate to expert panels when needed.
- Invest in skills: Train clinicians, laboratorians and data teams on AI fundamentals, data governance and clinical validation so models improve care instead of adding friction.
Why this approach works
AI2Health focuses on interpretable, biologically grounded methods, not black boxes. The goal is simple: turn complex data into decisions clinicians trust and public health teams can operationalize.
That starts with clear problem framing, rigorous validation, and workflows that respect privacy, safety and equity from day one.
Learn more
- Genomic variation basics from NHGRI: Structural variants
- Upskill your team with practical AI training by job role: Complete AI Training
Your membership also unlocks:
