AI | Personalizing Cancer Care with Artificial Intelligence

Artificial intelligence (AI) and big data are poised to revolutionize how we diagnose and treat cancer. Once limited to assisting drug development behind the scenes, AI is now enabling truly personalized care that could tailor treatment plans, dosages, and even cell and drug therapies to each individual patient. However, realizing this promise will require overcoming major regulatory hurdles to get these novel technologies safely into clinical practice.

AI is already being used in precision oncology, but mostly as an aid during drug development rather than at the bedside. Systems analyze patterns in vast biochemical and genomic datasets to discover new drug targets and biomarkers. Deep learning has improved neoantigen identification for personalized immunotherapies like CAR T-cell therapies. But until recently, true personalization focused only on doctors tailoring standard protocols case-by-case or developing customized autologous cell therapies from a patient’s modified cells.

This is changing rapidly as digital technologies increasingly power genuine personalization. AI analyzes medical images to improve cancer diagnoses compared to humans alone. Systems monitoring patients’ self-reported symptoms and quality of life have increased survival and reduced side effects in late-stage lung cancer. Such digital therapeutics automatically trigger alerts if symptoms exceed pre-defined thresholds, enabling automated personalized dosing adjustments overseen by doctors. As more patient data accumulates, rule-based systems are being replaced with machine learning models for better individualized care.

Digital technologies are even driving toward real-time personalization. Digital twins—dynamic virtual representations of patients integrating clinical, genomic, and physiological data streams—could simulate highly tailored diagnosis, treatment planning, and dose monitoring approaches. And generalist AI approaches that flexibly analyze any medical data without narrow specialization hold promise. Foundation models trained on huge healthcare datasets may interpret new information in an integrated manner critical for precision oncology and propose personalized combinatorial therapies.

These rapid advances pose huge regulatory challenges. Cell therapies, drugs and AI-based devices each face their own pathways. But emerging digital/AI-drug pairings defy traditional categories. Approval delays are exacerbating as gene and cell therapies flood regulators. Current frameworks cannot optimally regulate multidisciplinary concepts that merge AI, data, and new therapeutic modalities.

Several strategies could help regulators keep pace. The US FDA’s non-device designation exempts some clinical decision support tools from medical device requirements, turbocharging development. Independent AI testing platforms and simulation-based evaluations could augment traditional evidence pathways. Regulators are exploring predetermined change control plans allowing on-market updates to AI without reapproval. And “regulatory sandboxes” can safely trial innovative approaches.

However, transformative change is needed. Existing laws failed to anticipate integrated digital health concepts. Fragmentation between drug and device authorities causes contradictions and delays. Discrepancies between regions stall innovations accessible elsewhere. And precaution should not impede beneficially regulating emergent technologies according to risk-benefit profiles.

More agile thinking could solve these concerns. Pairing independent AI recommendations with physician oversight may safely fast-track personalized therapies before full evidence exists. Allowing concurrent approval of digital tools alongside drugs that rely on them could speed access. And oversight layers delineating professional responsibilities may regulate AI flexibly within multidisciplinary care teams.

Ethical issues also demand consideration. Patients should consent to AI’s role in their care and relationships. Systems must counter historic biases now amplified. Shared decision making between doctors, patients and AI may optimize personalized medicine’s patient-centered goals.

If regulators can resolve process, communication and philosophical challenges,AI promises to revolutionize cancer care. Digital technologies may design tailored cell therapies from a patient’s unique molecular profile in real-time. Combined with generalist AI’s holistic perspectives, this could optimize individualized treatment far beyond what any specialist could provide alone. With wisdom and agility, regulations can help deliver AI-powered precision medicine’s life-saving promise to all who need it.

Reference(s)

1. https://doi.org/10.1038/s41698-024-00517-w

Click TAGS to see related articles :

AI | MACHINE LEARNING | MEDICINE | ONCOLOGY