For years, DNA sequencing has made it possible to identify thousands of genetic variations in an individual, without knowing which ones were actually linked to a given disease. In 2025, a new generation of artificial intelligence tools is radically changing this situation. Researchers at the Icahn School of Medicine at Mount Sinai have developed a model capable not only of identifying potentially pathogenic mutations, but more importantly, of predicting the types of diseases associated with them, paving the way for a more interpretable and useful reading of the human genome1.
\n\n\n\nUntil recently, most predictive genetics tools were limited to classifying a mutation as benign or harmful. In practice, this left clinicians with lists of hundreds or even thousands of variants, without a clear hierarchy. New AI models are changing this approach by establishing probabilistic links between a genetic variation and specific clinical phenotypes, such as neurological disorders, immune diseases, or certain cancers. When tested on anonymized patient data, these systems are able to identify the mutation actually responsible for the disease among the top hypotheses, significantly reducing the time required for diagnosis1.
\n\n\n\nThis breakthrough is particularly significant for rare diseases, where the diagnostic journey can last several years. By 2025, more than 300 million people worldwide will be affected by a rare disease, the majority of which are genetic in origin2. By helping doctors identify relevant variants more quickly, AI reduces the number of unnecessary tests, accelerates care, and directs patients sooner to appropriate treatments or protocols. AI thus becomes a clinical decision-support tool, rather than merely a genetic calculation engine.
\n\n\n\nBeyond diagnosis, these models open up new avenues for medical research. By linking mutations to disease categories, AI enables the identification of genes, biological pathways, and mechanisms that have yet to be fully explored. For researchers and pharmaceutical companies, this means better prioritization of therapeutic targets, particularly in complex diseases where multiple genetic factors interact. According to an analysis published in Nature Reviews Drug Discovery, integrating AI into genomic analysis could increase the efficiency of the early stages of drug discovery by 20 to 30%3.
\n\n\n\nOne of the major contributions of these tools is the gradual shift from reactive medicine to predictive medicine. By anticipating the risks associated with certain genetic variations, it becomes possible to tailor medical monitoring, prevention, or treatment choices well before the first symptoms appear. This approach is fully in line with precision medicine, where medical decisions are based on individual biological profiles rather than statistical averages. By 2025, several hospitals will already be beginning to integrate this type of AI into their clinical genetics pathways4.
\n\n\n\nDespite these advances, researchers emphasize a key point: AI does not predict a fixed biological destiny. Predictions are based on probabilities and must be interpreted within a broader clinical context, taking into account the environment, lifestyle, and other non-genetic factors. AI acts as an intelligent filter, helping to prioritize relevant information, but the final decision remains a human one. This distinction is crucial to avoid any deterministic or anxiety-inducing misinterpretations regarding genetics.
\n\n\n\nThe rise of these technologies also raises major ethical questions. Who has access to these genetic predictions? How can we ensure the confidentiality of the most sensitive data imaginable? And how can we prevent this information from being used for discriminatory purposes, for example in insurance or employment? By 2025, healthcare institutions and regulatory authorities are emphasizing the need for strict oversight, combining algorithmic transparency, informed consent, and data governance5.
\n\n\n\nOur DNA speaks, and artificial intelligence is beginning to understand its language. This breakthrough is not only transforming genetics; it is redefining the way medicine approaches the complexity of life. By linking genomic data, diseases, and clinical decisions, AI is becoming a powerful bridge between biology and healthcare. The challenge in the coming years will not only be to improve the accuracy of these models, but to integrate them into healthcare systems in a responsible, humane, and equitable manner.
\n\n\n\nOn a related topic, check out our article “AI Identifies a Biomarker of Chronic Stress for the First Time Using Medical Imaging”, which shows how artificial intelligence models are opening up new avenues for understanding biological mechanisms and improving preventive healthcare.
\n\n\n\n\n\n\n\n1. Stein, D. et al. (2025). Expanding the utility of variant effect predictions with phenotype-specific models. Nature Communications.
\nhttps://www.nature.com \n
2. World Health Organization. (2024). Overview of rare diseases.
\nhttps://www.who.int \n
3. Nature Reviews Drug Discovery. (2024). Artificial intelligence in early-stage drug discovery.
\nhttps://www.nature.com \n
4. National Institutes of Health. (2024). AI in clinical genomics.
\nhttps://www.nih.gov \n
5. OECD. (2024). AI, genomics, and responsible innovation.
\nhttps://www.oecd.org \n
For years, DNA sequencing has made it possible to identify thousands of genetic variations in an individual, without knowing which ones were actually linked to a specific disease. In 2025, a new generation…
\n","protected":false},"author":2,"featured_media":483588,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[27],"tags":[59],"class_list":{"0":"post-483581","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-sante","8":"tag-parlonsia"},"yoast_head":"\n