Zhavoronkov and Mathematician Bud Mishra Describe Drugs as Forces Pushing the Body in Different Directions, Showing That Treatment Order Changes the Outcome
Zhavoronkov and Mathematician Bud Mishra Describe Drugs as Forces Pushing the Body in Different Directions, Showing That Treatment Order Changes the Outcome
On September 16, 2026, the journal Aging published a paper by Alex Zhavoronkov, founder and CEO of Insilico Medicine, and mathematician Bud Mishra of the Courant Institute at New York University. In their model, the body is a point in a space of possible biological states, and drugs and other interventions are forces pushing that point in different directions. Biological age is the minimum cost, in doses, risk, and time, required to return the body to a zone of normal function.
Existing theories of aging (López-Otín's hallmarks of aging, Aubrey de Grey's SENS, geroscience, Blagosklonny's hyperfunction theory) explain well what changes in cells and tissues with age. None of them answers the practical question facing a physician or drug developer: which intervention, at what dose, and in what sequence to apply to a specific organism right now. Six months ago, Zhavoronkov had already declared that aging clocks are dying and promised in their place not another number but a system that connects data to decisions. He now proposes to close this gap through control theory, the same branch of mathematics used to calculate the flight of aircraft and satellites. The calculations were formalized and verified by mathematician Bud Mishra. They state it in a single sentence:
Aging is a progressive loss of safe controllability; biological age is the minimum safe cost of restoring function.
The order of interventions itself changes the outcome, regardless of which interventions are chosen. This is a consequence of noncommutativity: applying A first and then B is not the same for the organism as the reverse order, and the authors compute the difference using the Lie bracket. Using a simplified model of an aged mouse liver, they compared three protocols: first a senolytic (clearing senescent "zombie" cells) followed by rejuvenating reprogramming; the reverse order; and both interventions simultaneously. The first order restores epigenetic integrity more effectively: the fewer senescent cells remain, the more strongly the reprogramming works. This is a calculation on a model with parameters drawn from the literature, and the authors describe the example as an illustration only.
To test the framework against data, the authors scored geroprotectors on eight parameters and published a top-20 ranking. At the top are caloric restriction and aerobic exercise, followed by diabetes drugs: the SGLT2 inhibitors dapagliflozin and empagliflozin, and the GLP-1 agonists liraglutide and semaglutide, all ranked above rapamycin, the most reproducibly validated geroprotector in mouse trials run by the Interventions Testing Program. Rapamycin loses places because of a penalty for immunosuppression and its narrow action on only one of the eight axes: the model predicts short courses at a specific life stage rather than continuous use.
The authors estimate that the current arsenal controls only a fifth of the organism's possible states and name the conditions for failure: the framework will be rejected if the cost of control does not outperform existing aging models and if predicted effects are not confirmed in independent trials. They describe the framework as a prioritization tool for the Insilico Medicine portfolio (the company where Zhavoronkov is CEO and Mishra serves as an advisor); the work had no external funding. The authors wrote this paper on AI-driven geroprotector discovery with the help of AI itself: Claude and GPT assisted with prose drafting and citation checking, while every equation and sentence was verified by the authors themselves, who take full responsibility.