Contraception, Vasectomy
Conditions
Keywords
vasectomy, ChatGPT
Brief summary
The investigators wish to perform a pilot study at the Manitoba Men's Health Clinic to assess if pre-vasectomy counseling with ChatGPT can safely streamline the consultation process by reducing visit times, while increasing patient satisfaction with the consultation process.
Detailed description
The release of ChatGPT, a natural language AI chat bot, has recently captured the attention of the general public and medical professionals world-wide. ChatGPT is a conversational AI language model designed by OpenAI that utilizes deep learning techniques to generate human like responses to questions. There has been increasing interest among the medical community on developing new ways to harness this technology to aid medical professionals. New applications have included writing scientific manuscripts, drafting correspondents, as well as testing its ability on standardized medical licensing examinations. ChatGPT was able to perform at or near passing threshold on the United States Medical Licensing Exam (USMLE). Despite ChatGPT demonstrating good medical knowledge, it is still unclear if this translates to real world clinical practice. Physician burnout is becoming an increasing problem. Per a recent Canadian Urologic Association (CUA) census, burnout was identified in 39% of urologists, with high number of patient visits/week to be cited as a predictor predictive of burnout. That begs the question, can ChatGPT be used for patient counseling and to improve patient flow in one's practice? ChatGPT has been shown to provide effective and safe responses to medical questions and even curbside consultations. However its ability to counsel patients in real time has not yet been studied. As such, the investigators wish to perform a pilot study at the Manitoba Men's Health Clinic to assess if pre-vasectomy counseling with ChatGPT can safely streamline the consultation process by reducing visit times, while increasing patient satisfaction with the consultation process. If the hypothesis provides true, further implementation of ChatGPT in a healthcare practice may work to reduce the burden on health care providers. There are no predicted adverse events of this study. There are no identified potential harms of this study.
Interventions
OTHERChatGPT Language Model
ChatGPT Language Model - ability to converse with the model and ask questions regarding vasectomies
Sponsors
University of Manitoba
Study design
Allocation
NON_RANDOMIZED
Intervention model
PARALLEL
Primary purpose
HEALTH_SERVICES_RESEARCH
Masking
SINGLE (Outcomes Assessor)
Eligibility
Sex/Gender
MALE
Age
18 Years to 65 Years
Healthy volunteers
Yes
Inclusion criteria
* Patients undergoing vasectomies interested in participating * Adult (\>18 years old) * Consenting
Exclusion criteria
* Do not consent, uninterested in participating
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Length of Consultation | Through study completion, estimated to be 2 months measured for each phone call | Length of pre-vasectomy consultation as measured from beginning to end of telephone call |
| Number of questions asked | Through study completion, estimated to be 2 months measured for each phone call | Number of questions asked during pre-vasectomy consultation |
| Satisfaction of patients assessed by Likert scale | Through study completion, estimated to be 2 months measured for each patient | Satisfaction of patients as measured by questionnaire with 10-point Likert Scale (\[0-10\] higher score correlates with higher satisfaction) |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Commonly asked questions | Through study completion period, estimated to be 2 months | Chat logs will be reviewed to examine common themes or questions and reported using descriptive statistics. |
| Incidence of inaccurate ChatGPT responses | Through study completion period, estimated to be 2 months | ChatGPT responses will be reviewed by two licensed clinicians experienced in vasectomies to determine rates of inaccurate responses by the language models responses to patient questions. Incidence of inaccurate ChatGPT responses will be reported and through descriptive analysis this will be described and inaccurate responses will be conveyed. |
Countries
Canada
Contacts
Primary ContactPremal Patel, MD
Outcome results
None listed