AI expands its role in medical genetics, raising both clinical promise and security risks

One of the most immediate impacts of AI in medical genetics lies in diagnosis, particularly for rare and complex genetic disorders. Patients with such conditions often face years of uncertainty as clinicians sift through vast numbers of genetic variants to identify the cause of disease. The review finds that AI-driven analysis platforms are now changing this process by ranking candidate genes using integrated evidence from genomic data, patient symptoms, disease databases, and scientific literature.

Artificial intelligence (AI) is rapidly redefining how genetic data are analyzed, interpreted, and applied in healthcare, according to a new review that charts the growing convergence between AI and medical genetics. Researchers argue that traditional analytical methods can no longer keep up, pushing AI from a supporting role into a central position in diagnosis, prediction, and treatment design.

The study, titled The Evolving Role of Artificial Intelligence in Medical Genetics: Advancing Healthcare, Research, and Biosafety Management,  published in the journal Genes, presents AI not as a single tool but as a foundational infrastructure shaping the future of precision medicine, while also introducing new ethical and security risks that regulators and clinicians must confront.

Together, the findings paint a picture of a healthcare system in transition. AI is shortening diagnostic delays for rare diseases, enabling non-invasive genetic insights from routine scans, accelerating drug discovery, and laying the groundwork for digital replicas of patients that could one day simulate disease progression and treatment response. At the same time, the authors warn that unchecked deployment could amplify bias, threaten data privacy, and lower barriers to misuse in genetic engineering.

AI transforms genetic diagnosis and disease prediction

One of the most immediate impacts of AI in medical genetics lies in diagnosis, particularly for rare and complex genetic disorders. Patients with such conditions often face years of uncertainty as clinicians sift through vast numbers of genetic variants to identify the cause of disease. The review finds that AI-driven analysis platforms are now changing this process by ranking candidate genes using integrated evidence from genomic data, patient symptoms, disease databases, and scientific literature.

These systems excel at managing scale. Whole exome and whole genome sequencing can generate millions of variants for a single patient, most of which are harmless. AI algorithms narrow this search space dramatically, allowing clinicians to focus on the most relevant findings. This has led to improved diagnostic yield and renewed progress in cases that previously remained unresolved despite extensive testing.

Beyond sequencing, AI is also reshaping how genetics connects with medical imaging. The review highlights the rise of imaging-genetics approaches that extract molecular and genetic information from standard pathology slides and radiological scans. By identifying subtle patterns invisible to the human eye, AI can infer tumor mutations, subtypes, and prognostic markers without the need for additional invasive procedures.

This shift has important equity implications. Advanced sequencing technologies remain costly and unevenly distributed worldwide. AI-based imaging analysis, by contrast, can leverage existing clinical data, potentially expanding access to molecular diagnostics in resource-limited settings. The authors argue that this could help narrow global disparities in cancer diagnosis and treatment if deployed responsibly.

Predictive medicine represents another major frontier. Traditional genetic risk scores rely on linear models that struggle to capture the complex interactions among thousands of genetic variants. Machine learning-based risk models outperform these older approaches by identifying non-linear relationships, improving prediction of conditions such as heart disease, diabetes, and neurodegenerative disorders. According to the review, this transition marks a broader shift in genetics from describing disease after onset to forecasting risk before symptoms appear, enabling earlier intervention and prevention.

From drug discovery to gene editing, AI becomes a generative force

While early applications of AI in genetics focused on analysis, the review highlights a more recent transformation: AI as a generative engine in therapeutic development. Instead of merely predicting outcomes, AI systems are now designing new drugs, proteins, antibodies, and gene-editing tools based on patterns learned from massive biological datasets.

In drug discovery, AI shortens timelines that traditionally stretched over a decade. By modeling molecular interactions, toxicity, and bioactivity, AI-driven platforms identify promising drug candidates and optimize their structures before laboratory testing begins. This approach reduces costs and failure rates, and has already produced therapies that moved into clinical testing faster than conventional methods allow.

Pharmacogenomics, which tailors drug choice and dosage to a patient’s genetic profile, also benefits from AI integration. The review notes that AI models combining genomic, transcriptomic, and protein data outperform single-data approaches in predicting drug response and adverse reactions. These tools support more precise prescribing, reducing trial-and-error medicine and minimizing harm.

Gene editing represents one of the most powerful and sensitive applications discussed. Technologies such as CRISPR hold promise for treating inherited diseases, but their safety depends on precise targeting. AI systems now play a critical role in designing guide RNAs that maximize on-target effects while minimizing unintended edits elsewhere in the genome. The review describes how AI-assisted design has already produced gene-editing tools with improved specificity, bringing clinical applications closer to reality.

At the same time, these advances introduce profound risks. The same generative capabilities that enable life-saving therapies could be misused to design harmful biological agents or evade existing safety controls. The authors frame this as a dual-use dilemma, warning that AI lowers technical barriers to sophisticated genetic manipulation. Addressing this threat requires safeguards embedded in AI systems, stronger oversight of genetic synthesis, and a culture of responsibility within the scientific community.

Digital twins, ethics, and the future of precision medicine

Looking ahead, the review identifies the digital twin concept as a potential unifying framework for AI in medical genetics. A digital twin would function as a continuously updated virtual replica of an individual, integrating genomic data, medical records, imaging, wearable sensors, and environmental factors. Such a model could simulate disease progression, test treatment options, and guide clinical decisions before interventions occur in the real world.

Although still largely aspirational, early implementations suggest significant promise. In fields such as cardiology, patient-specific digital models already help predict arrhythmia risk and inform surgical planning. The authors argue that extending this approach across medicine could fundamentally change how care is delivered, shifting from reactive treatment to proactive, personalized management.

Realizing this vision, however, depends on advances in multimodal AI systems capable of integrating diverse data types into a coherent model. It also requires robust governance frameworks to manage privacy, consent, and accountability. Genomic data are uniquely identifying and inherently linked to family members, making traditional anonymization insufficient. The review highlights emerging privacy-preserving methods, such as federated learning, that allow AI models to learn from data without centralizing sensitive information.

Bias remains another critical concern. Many genomic datasets disproportionately represent populations of European ancestry, raising the risk that AI tools perform poorly for underrepresented groups. Without intervention, the authors warn, AI could amplify existing health disparities rather than reduce them. Addressing this challenge requires both technical solutions and systemic changes, including more inclusive data collection and routine auditing of AI systems for fairness.