Bridging the boundaries between scientists and clinicians-mechanistic hypotheses and patient stories in risk assessment of drugs

Artificial intelligence methods for a Bayesian epistemology-powered evidence evaluation

RATIONALE, AIMS AND OBJECTIVES

The diversity of types of evidence (eg, case reports, animal studies and observational studies) makes the assessment of a drug's safety profile into a formidable challenge. While frequentist uncertain inference struggles in aggregating these signals, the more flexible Bayesian approaches seem better suited for this quest. Artificial Intelligence (AI) offers great promise to these approaches for information retrieval, decision support, and learning probabilities from data.

METHODS

E-Synthesis is a Bayesian framework for drug safety assessments built on philosophical principles and considerations. It aims to aggregate all the available information, in order to provide a Bayesian probability of a drug causing an adverse reaction. AI systems are being developed for evidence aggregation in medicine, which increasingly are automated.

RESULTS

We find that AI can help E-Synthesis with information retrieval, usability (graphical decision-making aids), learning Bayes factors from historical data, assessing quality of information and determining conditional probabilities for the so-called 'indicators' of causation for E-Synthesis. Vice versa, E-Synthesis offers a solid methodological basis for (semi-)automated evidence aggregation with AI systems.

CONCLUSIONS

Properly applied, AI can help the transition of philosophical principles and considerations concerning evidence aggregation for drug safety to a tool that can be used in practice.

Pharmacovigilance as Scientific Discovery: An Argument for Trans-Disciplinarity

Pharmacovigilance currently faces several unsolved challenges. Of particular importance are issues concerning how to ascertain, collect, confirm, and communicate the best evidence to assist the clinical choice for individual patients. Here, we propose that these practical challenges partially stem from deeper fundamental issues concerning the epistemology of pharmacovigilance. After reviewing some of the persistent challenges, recent measures, and suggestions in the current pharmacovigilance literature, we support the argument that the detection of potential adverse drug reactions ought to be seen as a serendipitous scientific discovery. We further take up recent innovations from the multidisciplinary field of serendipity research about the importance of networks, diversity of expertise, and plurality of methodological perspectives for cultivating serendipitous discovery. Following this discussion, we explore how pharmacovigilance could be systematized in a way that optimizes serendipitous discoveries of untargeted drug effects, emerging from the clinical application. Specifically, we argue for the promotion of a trans-disciplinary responsive network of scientists and stakeholders. Trans-disciplinarity includes extending the involvement of stakeholders beyond the regulatory community, integrating diverse methods and sources of evidence, and enhancing the ability of diverse groups to raise signals of harms that ought to be followed up by the network. Consequently, promoting a trans-disciplinary approach to pharmacovigilance is a long-term effort that requires structural changes in medical education, research, and enterprise. We suggest a number of such changes, discuss to what extent they are already in process, and indicate the advantages from both epistemological and ethical perspectives.

Causal Evidence and Dispositions in Medicine and Public Health

Since the introduction of evidence-based medicine, there have been discussions about the epistemic primacy of randomised controlled trials (RCTs) for establishing causality in medicine and public health. A growing movement within philosophy of science calls instead for evidential pluralism: that we need more than one single method to investigate health outcomes. How should such evidential pluralism look in practice? How useful are the various methods available for causal inquiry? Further, how should different types of causal evidence be evaluated? This paper proposes a constructive answer and introduces a framework aimed at supporting scientists in developing appropriate methodological approaches for exploring causality. We start from the philosophical tradition that highlights intrinsic properties (dispositions, causal powers or capacities) as essential features of causality. This abstract idea has wide methodological implications. The paper explains how different methods, such as lab experiments, case studies, N-of-1 trials, case control studies, cohort studies, RCTs and patient narratives, all have some strengths and some limitations for picking out intrinsic causal properties. We explain why considering philosophy of causality is crucial for evaluating causality in the health sciences. In our proposal, we combine the various methods in a temporal process, which could then take us from an observed phenomenon (e.g., a correlation) to a causal hypothesis and, finally, to improved theoretical knowledge.

What is the guidelines challenge? The CauseHealth perspective

This paper is an introduction to the conference, The Guidelines Challenge, held in Oxford in October 2017. My aim is to explain our motivation for organising this conference, as part of the research project Causation, Complexity, and Evidence in Health Sciences (CauseHealth). Depending on the professional starting point, the guidelines challenge can be interpreted in a number of ways. Our idea with this conference was to discuss guidelines from 3 overarching perspectives: practice, policy, and philosophy. In particular, we wanted to discuss some of the challenges that face anyone developing and implementing clinical guidelines in the evidence-based era of medicine. This introduction gives a brief overview of what CauseHealth sees as the guidelines challenge from these perspectives. More attention is given to the philosophical issues with which the CauseHealth project is particularly concerned, although a proper treatment or discussion of these issues naturally falls outside the scope of this introduction.