AI (Artificial Intelligence), Internship and Residency, Radiology
Conditions
Keywords
AI, Palestine, Intern Doctors, AI-reliance, Radiology
Brief summary
This study aims to enroll intern doctors and have them sit one of three identical radiology exams. The only difference between them is an AI-assistant. The differences between these groups will be used to measure the extent of AI reliance among intern doctors in Palestine.
Detailed description
This is a triple-arm trial investigating AI reliance in radiology among intern doctors in Palestine. The study will involve a radiology exam with three versions, a control, a sham AI (Correct answer) version, and a sham AI (incorrect answer) version. By comparing differences between the three groups, we aim to quantify AI reliance among this patient population.
Interventions
BEHAVIORALAI prompt (Correct)
This is a suggested answer in the guise of an AI assistant. The prompt was written by the authors and not an actual AI chat model. The suggested answer is correct.
BEHAVIORALAI prompt (Incorrect)
This is a suggested answer in the guise of an AI assistant. The prompt was written by the authors and not an actual AI chat model. The suggested answer is incorrect.
Sponsors
Al-Quds University
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
HEALTH_SERVICES_RESEARCH
Masking
QUADRUPLE (Subject, Caregiver, Investigator, Outcomes Assessor)
Masking description
The analyst will also be blinded.
Intervention model description
This study is a triple arm, triple blinded, parallel design randomized controlled trial. The study will measure how much intern doctors rely on AI assistance in radiologic interpretation and the behavioral impact of correct versus incorrect AI guidance. All interns will undergo a radiology exam with identical questions, and have their results compared across groups.
Eligibility
Sex/Gender
ALL
Healthy volunteers
Yes
Inclusion criteria
* Intern doctor in Palestine * Completion of at least 3 months from their 1 year internship * Confirmed prior training in radiologic interpretation
Exclusion criteria
* Does not consent to the study * Completion of the internship * Non-completion of at least 3 months of their 1 year internship
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| AI Reliance | Periprocedural | The extent of dependance of subjects on AI. It will be estimated based on a difference in mean score between the groups. We will also assess this outcome by creating an (AI-concordance field: for the intervention groups it will be how many times the subjects answered identically to the AI prompt, while for the control group it will be 0). AI reliance will be operationalized as: AI Reliance = Mean score improvement in the correct-AI group vs control Mean score decrement in the incorrect-AI group vs control We will compare the two different outcome measures to determine which better represents our outcome. |
| Exam time | Periprocedural | This will be defined as the length of time subjects spend completing the exam. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Correlation of baseline characteristics with AI reliance | Baseline | We will measure specific variables and their correlation with increased AI reliance. For this measure, we will depend on self-reported via a post-exam survey and include: gender, region, current clinical exposure, and current radiological exposure. We will then demonstrate the % of patients with the aforementioned characteristics and the differences in AI reliance in those aspects. |
| % of Subjects with a positive Perception of AI use in Radiology, and its correlation with AI reliance | Baseline | We will measure AI perception in radiology among subjects and its effect on their AI reliance. This will be done via a scale described in the literature, and by assessment of the % of subjects who have a positive, or negative outlook or perception on AI use in radiology. We will further test the relationship between AI reliance and AI perception. This will be done through the use of the scale described (Radiology Residents' Perceptions of Artificial Intelligence: Nationwide Cross-Sectional Survey Study) by Chen et al. |
| % of radiology interest as a specialty and its correlation with AI reliance | Baseline | We will measure radiology interest and its association with AI reliance. For this measure, we will use a validated tool for the measurement of radiology interest, described in the following study: "Assessing diagnostic radiology knowledge among Syrian medical undergraduates" We will then demonstrate the % of patients interested in specializing in radiology and the differences in AI reliance in those aspects. |
Countries
Palestinian Territories
Outcome results
None listed