All research

Artificial intelligence and the economics of clinical evidence (PhD research)

Parsa Kazemi · PhD in Economics, Management and Statistics, University of Messina (42nd cycle) ·

Research proposal · Status: Planning. Results are not shown until the work is finished.

Cover card: AI and the Economics of Clinical Evidence

In 2026 I was admitted, first in the ranking, to the PhD in Economics, Management and Statistics at the University of Messina (42nd cycle). I work in the AI-HealthLab of the Department of Mathematical and Computer Sciences, Physical Sciences and Earth Sciences. The programme is an executive PhD with Biofuture Medicine.

The question

Developing a new medicine costs billions and takes more than ten years. A large part of the cost sits in the clinical trial, and especially in its control arm, where people are enrolled mainly to show what would have happened without treatment. Generative AI can now simulate how a disease would progress in one person. Such a simulated person is a digital twin. If twins are good enough, they could stand in for part of a control group.

My research treats this as an economic question, not only a technical one. Predicting who progresses fast is a prediction problem. Deciding which patients to study, and when simulation is cheaper and safer than recruiting, is a question about costs, value and risk.

Three research questions

  1. Clinical validity. Can twins trained on a small, well-described cohort reproduce plausible untreated trajectories?
  2. Trial efficiency. How much trial cost and time can be saved by replacing part of a control group, once the cost of the pipeline and the risk that regulators say no are counted?
  3. Decision threshold. Under which conditions is a twin-based design better than normal recruitment? I plan to use causal machine learning (double machine learning and causal forests) and decision analysis.

Plan

  • Year 1. Literature review of the three research directions, access to longitudinal datasets, the economic framework, and an evaluated twin pipeline.
  • Year 2. The trial-efficiency analysis and the start of the threshold analysis, plus a period of research abroad.
  • Year 3. Synthesis, thesis and journal articles.

Where I am now

I am in year 1. I wrote two literature reports on the field and on data sources, and a design note for the economic model. Data access and the next tasks are being agreed with my academic team. No results exist yet, and this page will show them only when they do.

phd · digital twins · health economics