Daphne Koller: Biomedicine and Machine Learning

Koller resists absolute predictions about curing every major disease or extending life indefinitely. Her argument is not that such outcomes are impossible, but that calling them inevitable would ignore how much damage may already have occurred by the time a disease is detected. A genuine cure could require restoring damaged parts of the body to their original state, while current medicine more often manages disease than eliminates it.

The level of understanding varies sharply by disease. Koller places some diseases in the 70–80 range, but says Alzheimer’s disease is closer to zero because its hypotheses remain insufficiently validated. She also argues that Alzheimer’s, schizophrenia, and type 2 diabetes should not necessarily be treated as single diseases: they may be heterogeneous collections of mechanisms that produce similar clinical outcomes. The breast-cancer analogy illustrates why identifying molecular subtypes must precede a complete mechanistic account.

Longevity and disease overlap without being identical. Disease risk rises with age for many non-childhood diseases, while aging also produces changes not tied to one specific disease. Koller describes accumulated DNA damage, misfolded proteins, and inflammation as cellular wear that may contribute to disease, but she frames a longer health span—not immortality—as the more practical aspiration. Her example is a life of roughly 120 years in good health, while acknowledging that the target age remains debatable.