Large Linguistic Model for Clinical Reaoning of Physical Therapy Students

Feasibility of a Randomized Controlled Trial of Large Artificial Intelligence-Based Linguistic Models for Clinical Reasoning Training of Physical Therapy Students. A Randomized Controlled Trial

Clinical reasoning is a fundamental skill for physical therapy students, enabling them to collect and interpret patient information to make accurate diagnoses and treatment decisions. Traditional training methods often limit students' exposure to a diverse range of clinical cases, which can restrict the development of these skills. The integration of Large Language Models (LLMs), such as ChatGPT, into physical therapy education offers a novel approach to enhance clinical reasoning by simulating interactive and realistic patient scenarios.

This randomized controlled trial aims to evaluate the effectiveness of an LLM-based educational intervention in improving clinical reasoning skills in physical therapy students. The study will recruit a total of 200 third-year physiotherapy students from multiple university institutions. Participants will be randomly assigned to one of two groups:

1. Experimental Group - Students will receive LLM-based training, engaging with a conversational artificial intelligence model to solve clinical cases over an 8-week period. The model will provide real-time responses to their questions, allowing them to refine their diagnostic and treatment reasoning.
2. Control Group - Students will follow the standard curriculum, participating in conventional case-based learning and supervised clinical reasoning exercises without AI-based assistance.

The primary outcome of the study is the improvement in clinical reasoning skills, assessed through standardized written case evaluations and structured practical examinations. Secondary outcomes include changes in digital competence, student engagement levels, overall satisfaction with the educational approach, and cost-effectiveness of the intervention.

By assessing the impact of LLMs on clinical reasoning training, this study seeks to determine whether AI-driven educational tools can effectively complement traditional physiotherapy education and improve student preparedness for real-world clinical practice.

Study Overview

• Status: Recruiting
• Conditions: Healthy; Artificial Intelligence (AI)
• Intervention / Treatment: Other: Large Language Model; Other: Conventional

Detailed Description

Clinical reasoning is a key competency for physical therapy students, allowing them to assess, diagnose, and create treatment plans based on patient information. Despite its importance, traditional educational approaches often limit students' exposure to a broad variety of clinical cases, restricting their ability to develop comprehensive reasoning skills. Advances in artificial intelligence, particularly Large Language Models (LLMs) such as ChatGPT, offer a promising solution by simulating realistic and interactive clinical scenarios.

This randomized controlled trial (RCT) aims to evaluate the effectiveness of an LLM-based intervention compared to traditional training methods in improving clinical reasoning skills among physical therapy students. The third-year students will be randomly assigned to either the experimental group, receiving AI-driven case-based training, or the control group, following conventional curriculum-based case discussions.

The intervention will last 8 weeks, during which students in the experimental group will interact with an LLM to solve weekly clinical cases, mimicking real-world patient encounters. The model will function as a virtual patient, responding to students' inquiries and allowing them to refine their diagnostic reasoning and treatment planning. In contrast, the control group will participate in traditional written and tutor-led case discussions.

Statistical Analysis Plan

Data will be analyzed using SPSS version 29.0 (SPSS Inc., Chicago, IL, USA). Descriptive statistics will be used to summarize baseline characteristics of participants, with continuous variables expressed as mean ± standard deviation (SD) or median [interquartile range], depending on normality, and categorical variables presented as frequency (n) and percentage (%). Normality of distributions will be assessed using the Kolmogorov-Smirnov test and Shapiro-Wilk test. Between-group comparisons will be performed using; Independent t-tests or Mann-Whitney U tests for continuous variables; Chi-square tests or Fisher's exact test for categorical variables; Repeated-measures ANOVA or linear mixed models will be used to evaluate changes over time in clinical reasoning scores, digital competence, and satisfaction levels.

Logistic regression models will be applied to explore predictors of engagement with the LLM-based intervention. Effect sizes (Cohen's d, Rosenthal's r) will be calculated to measure the magnitude of differences observed. A cost-effectiveness analysis will be conducted by comparing the cost of implementing the LLM-based intervention with the improvement in clinical reasoning scores and student engagement levels.

Statistical significance will be set at p < 0.05, and all analyses will be conducted using a two-tailed approach.

• Study Type: Interventional
• Enrollment (Estimated): 60
• Phase: Phase 2

Contacts and Locations

• Study Contact: Name: Alfredo Lerín Calvo, Professor; Phone Number: +34620187457; Email: alfredo.lerin@lasallecampus.es

Study Locations

• Spain
  • Madrid
    • Madrid, Madrid, Spain, 28023
      • Recruiting
      • Centro Superior de Estudios Universitarios La Salle
      • Contact: Alfredo Lerín Calvo, Professor; Phone Number: +34620187457; Email: alfredo.lerin@lasallecampus.es
      • Principal Investigator: Raúl Ferrer Peña, PhD

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult
• Accepts Healthy Volunteers: Yes

Description

Inclusion Criteria:

• Students enrolled in the third year of the Physiotherapy program at La Salle Centre for Higher University Studies (LCHUS)
• Participants must be between 18 and 30 years old.
• Students must agree to participate in the study by signing an informed consent form after being briefed about the study's objectives, procedures, and potential risks.
• Participants must be willing to engage with the LLM-based platform (for the experimental group) or participate in traditional learning activities (for the control group) for the duration of the study.

Exclusion Criteria:

• Students with previous clinical experience beyond the third year of physiotherapy education.
• Physical or cognitive disabilities that may interfere with the ability to participate in or benefit from the intervention (e.g., vision, hearing, or motor impairments).
• Students who do not provide informed consent to participate in the study.
• Students who do not possess sufficient proficiency in Spanish or English to understand the materials and the intervention.

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Treatment
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Double

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: LLM group; Participants in the experimental group will undergo an 8-week intervention incorporating Large Language Model (LLM)-based training into their clinical reasoning education. Students will engage in weekly clinical case simulations using an LLM-powered platform (ChatGPT), where they will interact with the model to obtain patient information, formulate diagnoses, and propose treatment plans. The LLM will provide real-time responses, simulating a virtual patient encounter. The training will complement the standard curriculum, allowing students to practice clinical reasoning skills in a structured and interactive AI-assisted environment. At the end of the intervention, participants will complete a final case-based assessment to evaluate improvements in clinical reasoning, digital competence, and engagement with the technology.	Other: Large Language Model; The intervention in the experimental group is distinguished by the integration of a Large Language Model (LLM)-based interactive platform (ChatGPT) into clinical reasoning training for physical therapy students. Unlike traditional educational approaches, this intervention provides real-time, AI-generated patient interactions, allowing students to actively engage in virtual clinical case simulations.
Active Comparator: Conventional learning group; Participants in the control group will follow the standard curriculum for clinical reasoning training over an 8-week period, without exposure to the LLM-based intervention. Students will engage in weekly case-based discussions using traditional learning methods, including written case analyses and supervised discussions with instructors. These sessions will follow the usual educational framework used in physical therapy training programs, emphasizing diagnostic reasoning and treatment planning through instructor-led guidance. At the end of the training period, participants will complete a final case-based assessment to evaluate their clinical reasoning skills, digital competence, and overall engagement with the learning process.	Other: Conventional; The intervention in the control group follows a traditional case-based learning approach, which is commonly used in physical therapy education. Unlike the experimental group, this training method relies solely on human-led instruction and written case analysis, without the integration of artificial intelligence or interactive digital tools.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Clinical Reasoning Performance	This outcome measures the improvement in students' clinical reasoning skills after the intervention. Students will be assessed based on their ability to collect, interpret, and analyze patient information, and formulate accurate diagnoses and treatment plans. This will be evaluated through both written case studies and practical exams using the Lasater rubric, being this scale the instrument used for evaluating this outcome.	Assessed at the beginning and end of the 8-week intervention through case-based assessments and practical evaluations.

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Student engagement with the intervention	This outcome measures the level of active engagement of students with the LLM-based platform (experimental group) and the traditional case-based learning (control group). Engagement will be assessed based on the frequency of interactions, duration of usage, and completion rates of the assigned cases.	Monitored throughout the 8-week intervention period with weekly tracking of student interactions and case completions.
Satisfaction with the educational approach	This outcome assesses overall student satisfaction with their learning experience, focusing on the effectiveness, usability, and perceived value of the intervention. Satisfaction will be measured using a Visual Analog Scale (VAS), where students rate their level of satisfaction with the training method.	Calculated at the end of the intervention period, using the costs associated with providing access to the LLM-based platform and comparing it to the improvements observed in other outcomes.
Digital competences	This outcome evaluates the students' ability to effectively use digital tools, particularly AI-driven platforms like ChatGPT, in the context of clinical reasoning. Students will complete a digital competence questionnaire assessing their skills across several domains, such as data management, health communication, and digital content creation. It will be assessed with the Digital competences questionnaire developed by Montero-Delgado et al. (2020)	Evaluated at the start and end of the 8-week intervention via the ad hoc digital competence questionnaire.

Collaborators and Investigators

• Sponsor: Neuron, Spain

Publications and helpful links

General Publications

• Lasater K. Clinical judgment development: using simulation to create an assessment rubric. J Nurs Educ. 2007 Nov;46(11):496-503. doi: 10.3928/01484834-20071101-04.
• Milad D, Antaki F, Milad J, Farah A, Khairy T, Mikhail D, Giguere CE, Touma S, Bernstein A, Szigiato AA, Nayman T, Mullie GA, Duval R. Assessing the medical reasoning skills of GPT-4 in complex ophthalmology cases. Br J Ophthalmol. 2024 Sep 20;108(10):1398-1405. doi: 10.1136/bjo-2023-325053.

Study record dates

Study Major Dates

• Study Start (Actual): September 1, 2023
• Primary Completion (Estimated): June 30, 2026
• Study Completion (Estimated): July 30, 2026

Study Registration Dates

• First Submitted: January 30, 2025
• First Submitted That Met QC Criteria: February 4, 2025
• First Posted (Actual): February 5, 2025

Study Record Updates

• Last Update Posted (Estimated): December 4, 2025
• Last Update Submitted That Met QC Criteria: December 3, 2025
• Last Verified: February 1, 2025

More Information

Terms related to this study

• Keywords: Artificial Intelligence; Clinical reasoning; Large Language Models; Physical Therapy education
• Additional Relevant MeSH Terms: Organizations; Health Care Economics and Organizations; Machine Learning; Artificial Intelligence; Algorithms; Mathematical Concepts; Neural Networks, Computer; Deep Learning; Congresses as Topic; Large Language Models
• Other Study ID Numbers: ALC001NR005

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

IPD Plan Description

In this study, the individual participant data (IPD) will not be shared publicly. The data collected, including clinical reasoning assessments, digital competence evaluations, and satisfaction scores, will be used solely for the purposes of this research study. Access to participant data will be restricted to the research team and will not be made available for sharing with external parties.

The privacy and confidentiality of the participants will be strictly maintained throughout the study. All data will be anonymized, and any identifiable information will be securely stored and protected in compliance with applicable data protection regulations. Additionally, the study's results will be shared in aggregate form, ensuring that no individual's data is disclosed or identifiable in any public reports or publications.

As part of the ethical commitment to safeguarding participant privacy, any requests for access to IPD will not be granted.

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No