APTA of NH 2025 Presentation - Innovation in Evidence-Based Practice
Bayesian Reasoning Bridges the Gap Between Research and Practice
Just a post to demonstrate how thinking and writing, and then writing and thinking, begets other opportunities. Having put more time than I expected into the Stats4PT lessons on Bayesian reasoning (lessons three, four and five), I felt well prepared to put together a proposal to speak at the 2025 APTA of NH conference this Fall (October 18) in Manchester at Franklin Pierce University. This will be my 3rd presentation at this conference in 4 years. But this is the first that is so squarely centered on the work I’m actively engaged in doing.
I’m posting here for any readers pondering the switch to academia. Once you’re thinking and writing about a question or problem; and then writing and thinking about that question or problem, you are doing scholarship. And once you start doing scholarship - you have the opportunity to share it. The more places you share it - the more you engage with it and get others to engage with what you’re working on. You refine your thinking, with the goal of refining all of those that take the time to engage with you.
I regularly talk with clinicians with a DPT that are interested in entering academia. That’s wonderful. An important part of that conversation is that the DPT is not the traditional path to academia. It’s the path to the clinic. To transition to academia you must be a scholar. How you get there is up to you. Whether you pursue a traditional academic doctoral degree (PhD, EdD, ScD) or not depends on what sort of scholar you’d like to be. However, regardless of whether you earn one these terminal (scholar producing) degrees or not, you must be a scholar to enter (and remain in) academia.
Anyway - Here’s the proposal I submitted - once I have the presentation completed, I’ll also share it here. I thought this may be a good example of how something I was working on (teaching, thinking, writing substack posts), was turned into an opportunity for a “peer-reviewed dissemination” (the marker in academia of having done scholarship.
Innovation in Evidence-Based Practice:
Bayesian Reasoning Bridges the Gap Between Research and Practice
Course Description
This session presents a timely and transformative re-examination of evidence-based practice (EBP) through the lens of Bayesian reasoning—an approach that, while foundational to clinical decision-making, remains largely absent from how EBP is taught, applied, and measured. For physical therapists striving to integrate research evidence into individualized care, Bayesian reasoning offers a missing link between population-level data and patient-specific decisions.
At its core, Bayes’ Theorem—often called the “knowledge equation”—describes how to update the probability of a hypothesis (or cause) in light of new evidence (or effects). While traditionally used in statistical modeling, this theorem also captures the exact reasoning process clinicians use every day: refining diagnoses, updating prognoses, and selecting interventions based on evolving patient information. Yet despite this alignment, EBP has historically been tethered to frequentist statistics, emphasizing p-values and binary thresholds that often contradict the fluid and recursive nature of clinical care.
To borrow an analogy from author David Foster Wallace, it’s like the story of an older fish swimming by two younger fish and saying, “Morning, boys, how’s the water?” The two younger fish swim on for a bit before one turns to the other and asks, “What the hell is water?” Clinicians are often like those fish—immersed in Bayesian reasoning without realizing it. They update beliefs with new information, weigh risks and likelihoods, and adjust decisions on the fly, all while unaware that this intuitive process reflects the structure of Bayes’ Theorem.
This session will demonstrate how Bayesian inference provides a coherent framework that aligns with the actual logic of clinical reasoning, offering a structured and justifiable way to incorporate both prior knowledge and new evidence into practice. Participants will explore Bayesian methods as tools for:
Estimating the probability of diagnoses from symptoms and signs (abductive reasoning),
Evaluating treatment effectiveness in individual patients (deductive application of research),
Adjusting prognostic expectations over time (recursive updating), and
Bridging the divide between what research says “works on average” and what’s most likely to work for a specific patient.
We will explore real-world clinical examples and research scenarios to illustrate how Bayesian reasoning reshapes both the development of knowledge and its application at the point of care. Emphasis will be placed on:
How diagnostic reasoning, treatment planning, and prognosis estimation are inherently Bayesian processes.
Why failing to acknowledge Bayes’ Theorem in research and education has created a disconnect between evidence generation and clinical use.
How Bayesian frameworks allow us to interpret diagnostic test properties, treatment likelihoods, and outcome predictors in a way that maps onto how clinicians actually think.
The role of Bayesian reasoning in the emerging field of causal modeling and critical realist reviews.
This session is not just about teaching a different statistical method—it is about helping clinicians and educators see the underlying logic of their own reasoning more clearly, and to align evidence-based practice with clinical reality.
Bayesian reasoning is not an advanced statistical add-on—it is the knowledge core of rational clinical inference. Embracing it can transform how we teach, evaluate, and perform EBP. For educators, it provides a clearer way to teach inference and justify action. For researchers, it offers a way to generate findings that translate more directly into practice. And for clinicians, it legitimizes and strengthens the way we already think—bringing clarity to uncertainty and structure to judgment.
This session is ideal for clinicians, educators, and researchers who are eager to close the gap between research and practice—not by working harder within the current paradigm, but by stepping into a new one.
Session Objectives:
By the end of this session, participants will be able to:
Explain Bayes’ Theorem and how it serves as the foundational logic for clinical decision-making under uncertainty.
Contrast frequentist and Bayesian approaches to statistical inference, highlighting why Bayesian reasoning better mirrors real-world clinical reasoning.
Apply Bayesian thinking to diagnostic, prognostic, and treatment decisions in physical therapy practice.
Recognize how prior beliefs, likelihoods, and new patient information interact to update clinical judgments.
Reflect on the ways Bayesian reasoning is already implicitly used in clinical workflows—and how making it explicit can improve evidence translation and patient outcomes.
References
Kyrimi E, Dube K, Fenton N, et al. Bayesian networks in healthcare: What is preventing their adoption? Artificial Intelligence in Medicine. 2021;116:102079. doi:10.1016/j.artmed.2021.102079
Herrle SR, Corbett EC, Fagan MJ, Moore CG, Elnicki DM. Bayesʼ Theorem and the Physical Examination: Probability Assessment and Diagnostic Decision Making: Academic Medicine. 2011;86(5):618-627. doi:10.1097/ACM.0b013e318212eb00
Lohse K. In Defense of Hypothesis Testing: A Response to the Joint Editorial from the International Society of Physiotherapy Journal Editors on Statistical Inference through Estimation. Physical Therapy. Published online September 7, 2022:pzac118. doi:10.1093/ptj/pzac118
Yu H, Moharil J, Blair RH. BayesNetBP: An R Package for Probabilistic Reasoning in Bayesian Networks. Journal of Statistical Software. 2020;94(3). doi:10.18637/jss.v094.i03
Koponen V. Conditional probability logic, lifted Bayesian networks, and almost sure quantifier elimination. Theoretical Computer Science. 2020;848:1-27.
Gelman A, Vehtari A. What are the Most Important Statistical Ideas of the Past 50 Years? Journal of the American Statistical Association. 2021;116(536):2087-2097. doi:10.1080/01621459.2021.1938081
Collins S. Welcome to Stats4PT: Rethinking statistical reasoning in physical therapy. PeripateticPT. Published 2024. Accessed April 23, 2025. https://peripateticpt.substack.com/p/welcome-to-stats4pt
Speaker Bio: Dr. Sean M. Collins, PT, ScD
Dr. Sean Collins is Professor of Clinical Inquiry in the Department of Physical Therapy at Plymouth State University, where he teaches a three-course sequence on clinical reasoning, research methods, and statistical inference in practice. With over 25 years in academic leadership and clinical research, Dr. Collins’ career has centered on bridging evidence and clinical practice through innovative and interdisciplinary approaches.
He holds a BS in Exercise Physiology, an MS in Physical Therapy, and an ScD in Work Environment (with a focus in human factors, ergonomics, and epidemiology), all from the University of Massachusetts Lowell. Dr. Collins was the founding Program Director of PSU’s DPT program and previously served as Department Chair and Program Director at UMass Lowell.
Dr. Collins’ work emphasizes the application of Bayesian reasoning, causal inference, and decision-making under uncertainty in clinical education and practice. He is the creator of Stats4PT, a Substack-based learning platform that translates complex statistical and epistemological concepts into clinician-friendly formats, where his ongoing series on Bayesian reasoning is redefining the relationship between research and practice.
His research has been published in Physical Therapy, Cardiopulmonary Physical Therapy Journal, BMJ Open, and American Journal of Industrial Medicine, among others. He served as Editor-in-Chief of the Cardiopulmonary Physical Therapy Journal from 2016 to 2021 and received the Linda Crane Lecture Award in 2018 from the APTA.
Dr. Collins continues to advocate for advancing evidence-based practice through integration of Bayesian approaches that reflect how clinicians naturally reason and make decisions in the context of patient care. He also supports clinical decision-making using modeling, simulation, and game-based learning.
Other Comments
Although this session explores the statistical and philosophical underpinnings of clinical reasoning, its aim is deeply practical: to improve everyday clinical decision-making by offering a more accurate and intuitive framework for applying evidence in real-world contexts. The presentation will connect abstract concepts like Bayes’ Theorem to highly relatable clinical scenarios, making Bayesian reasoning immediately accessible and useful. This session fulfills the conference’s invitation for innovation by re imagining how evidence-based practice is taught and applied—bridging the persistent gap between research and practice. Attendees will leave with tools to reason more clearly, make better-informed clinical decisions, and understand why traditional EBP often falls short in practice.