Controlling and Utilizing Uncertainty in the Health Sciences (P2022RT4AT)
Uncertainty threatens the reliability of scientific results and inferences in the health sciences as well as the efficacy of public health policies. Uncertainty arises from the awareness that different flaws may affect evidence sampling, formalization, and reporting. The methodological literature offers a rich variety of instruments to address many sorts of uncertainty affecting scientific inference. These tools are extremely important and indispensable; yet, it is precisely because of the quantity, variety, and sophistication of these instruments, and because of their heterogeneous theoretical bases, that a foundational reflection on this sort of topic may considerably contribute to their correct implementation and wider use in policy-making and decision making in general.

Philosophy of science and epistemology have been long contributing to this theme comprehensively. These contributions may help science and methodology to progress in three ways:

1. Provide the rationale for the “correctness” or “truth-conduciveness” of the methodological techniques developed so far, in order to track and measure uncertainty.

2. Help develop additional tools for types of uncertainty which have not yet been “operationalized” by methodologists.

3. Build a comprehensive framework where such tools can be combined and put them into perspective (a taxonomy of uncertainty for a taxonomy of methods to track it).

Recent developments in formal epistemology and logic provide promising methods to address these issues. They enable one to represent the components of a scientific model, to determine its causal relations, and to operate formally on it, isolating the sources of uncertainty and weighing the impact of critical factors (biases, errors, etc.). The goal of this project is to draw on these developments in order to:

1. Analyze and categorize the different sources of uncertainty in the health sciences.

2. Draw on recent advances in Bayesian epistemology, causal modeling, logic, and AI to make these sources explicit and technically controllable.

3. Systematically explore the role of these sources of uncertainty for confirmation, further evolve these models, and develop new strategies to make scientific inferences more reliable.

Benefits: On the theoretical side, the project promises to improve the accuracy of our interpretation of scientific data and of the resulting models. On the practical side, having developed a principled way to minimize or utilize uncertainty in the interpretation of data, will also provide a rigorous method to reduce the risks in developing health policies based on data and results that are usually connotated by severe epistemic uncertainty. To make these results practically applicable, we develop a prototype software that allows scientists and policymakers to input known and unknown sources of uncertainty and to compute their effects on the confirmatory impact of the available evidence.

This PRIN project is a cooperation of the CPSP and the Center for Logic, Language and Cognition (LLC) at the University of Turin.


Philosophy of Pharmacology: Safety, Statistical Standards and Evidence Amalgamation (ERC GA: 639276)
What is the nature of causality and evidence in pharmacology? On what grounds is causal assessment licensed in pharmacology? Do current evidence standards do justice to the purpose of risk prevention and safety? These are some of the questions behind the ERC-funded project Philosophy of Pharmacology: Safety, Statistical Standards, and Evidence amalgamation. The project has three objectives:

1. To provide a foundational analysis on statistical/causal inference with a focus on the critical assessment of current practices in drug approval and pharmacosurveillance.

2. To build a unified epistemic framework within which different kinds of evidence for pharmaceutical harm can be combined and used for decision: evidence amalgamation.

3. To provide a theoretical framework for the development of new standards of drug evaluation.

The project addresses these issues by advancing a system for evidence amalgamation for the purpose of probabilistic causal assessment; i.e. where rather than grading the evidence, we grade the hypothesis of the causal link. Although the system integrates various pieces of evidence through an epistemic Bayesian network, it is intended to incorporate any kind of statistical data (e.g. also results from classical Fisherian or Neyman-Pearson tests), or any kind of studies (from laboratory experiments on certain cell cultures, to animal studies, up to machine learning analysis of “big data”).


The problem of collecting, analyzing and evaluating evidence for causal assessment is a central one in science-based policy. In particular, due to the ongoing decision-making process in policy-making, and the radical uncertainty affecting it, probabilistic tools of causal assessment are essential to allow flexible yet justified and robust inferential procedures. 

The ERC-funded PhilPharm project focused on probabilistic causal assessment in pharmacovigilance. This constituted a formidable laboratory for the integration of causal (inference) theories, statistical paradigms, scientific methodology and epistemology.
We aim to develop E-Synthesis in further directions both “horizontally” (by applying it to other fields in policy making and more generally in science-based decision-making) and “vertically” (by further investigating its foundational implications and mathematical modelling). For this, we aim to gather scholarly expertise in probability and probabilistic reasoning, formal epistemology, rational choice theory, philosophy of science and scientific methodology, epistemology, sociology of science, science policy. We also aim to develop an open-source application integrating expert systems approaches of diverse kinds (Bayesian expert systems, AI, Machine Learning) in collaboration with possible end-users (e.g. Governmental Agencies, Policy-makers, scientific experts). 
The project will aim to:

1. further investigate the foundational bases underpinning E-Synthesis (epistemic dynamics, probability kinematics, justification of causal inference, statistical data analysis, methodological issues and exogenous factors impacting on evidence quality). In particular, strategic dimensions in evidence sampling, analysis, disclosure and communication will be in focus (see Osimani, 2024);

2. further develop the E-Synthesis framework and implement it into a prototype software for decision support. As an immediate goal, we privilege the development of a decision aid for pharmacosuveillance, however, any other relevant field of application is considered (especially environmental policy). 

More detailed project description