The true meaning of personalized medicine.

I am not my genome. You cannot understand my desires or preferences, the trajectory of my life, nor my plans for the future from my DNA. Yes, in time more of what will be offered to treat me with will be shaped by my molecular biology, and for that I am profoundly thankful. But personalized medicine is not defined in the base pairs of my double helix, nor in the methylation pattern of my epigenome. It is defined, as it always was, by my choices and my needs. Somehow we have allowed the idea of personalized medicine to be colonized by molecular biology, by the promise of therapeutic magic bullets [1, 2]. The promise is that a molecular understanding of individual variation will lead to targeted molecular therapies. Personalization is reduced to a molecular match with the individual, to a crafting of a molecule rather than an engagement with a person. Modern medicine is sometimes criticized for inventing diseases that can then be treated by new medicines, or leaning too heavily on drug therapy when a more holistic assessment of patients would lead to non-pharmacological interventions. Consider for example the choice between prescribing a cholesterol lowering agent, or asking a patient to exercise, change their diet and lose weight. The pill is easier to prescribe than embarking on the journey of partnering with an individual to realign the way they conduct their lives, and the reasons that shape their behaviors. Personalized molecular medicine can similarly be criticized for diverting our attention from the traditional meaning of personalized medicine, where the care of an individual is shaped by an understanding of their entire context, and an engagement with their wishes and needs. There will, we all hope, come a day when much of medicine is bespoke, and medicines are purpose-built an atom at a time for the individual. But even when that happens, there will be other choices that need to be made, trade-offs to be considered; there will be preferences and decisions about care that have nothing to do with molecules or survival curves. The choice to be treated, or how to be treated, is much more complicated than that. Consider for example, the diff iculty in making end-of-life decisions. Classical decision making theory is often derided for being too abstract, but at its core, the notions of individual preference and utility are quantif ied, and the equation for making a decision is truly personal [3]. So, no one is saying the grand endeavor of molecular medicine has failed to deliver, or that is misguided. It is however time to reclaim the meaning of personalized medicine in the name of the whole individual, and not just our molecular biology. Informatics is not blameless in the retreat from the personal. In the early days of AI in medicine, generic models of clinical knowledge were soon replaced by the idea of patient-specific models, where knowledge about diagnosis and treatment was adapted and modif ied computationally to create models about the individual [4]. That notion has somehow been lost. Indeed the technologies that informatics concerns itself with have in the main moved in the opposite direction, searching for generic instead of individual approaches.