Mohs Chatbot Instructions Supplementary Material
Description
The study hypothesized that an interactive chatbot, programmed using the Landbot no-code platform, would improve patient understanding of postoperative care instructions after Mohs micrographic surgery (MMS) compared to standard written instructions. The chatbot was designed to deliver information via multiple-choice prompts and hardcoded if-then logic, potentially enhancing engagement and retention without relying on AI. The data consists of quantitative results from a randomized controlled study with 50 English-speaking adult MMS patients (≥18 years) at the Desai Sethi Medical Center, University of Miami. Participants were divided into two groups: 25 received standard written instructions (control), and 25 used the chatbot (intervention). Data was gathered post-surgery via a 10-question multiple-choice comprehension quiz testing knowledge of wound care, activity restrictions, infection prevention, and emergencies (scored as percentage correct) and secondary outcomes assessed with 0–100 sliders for ease of understanding, ease of use, satisfaction, confidence in wound care, and likelihood of contacting staff. Data collection followed IRB approval and informed consent. Instructions were authored by board-certified surgeons and mirrored in both formats. The chatbot used hardcoded flows based on uploaded instructions, with no patient data input to ensure ethical compliance. Supplementary materials include: Comprehension Questionnaire: The 10 questions with answers and scoring rubric Chatbot Flow Diagram: Visual of the decision-tree structure. Use this to understand chatbot navigation. Postoperative Instructions: Full text covering daily wound care (e.g., "Clean with gentle soap and water"), restrictions (e.g., "No swimming"), bruising/bleeding management, and infection signs. This is the content used in both written and chatbot formats. The chatbot group scored 72.0% on comprehension (vs. 64.4% control), but the difference was not statistically significant (p=0.10, unpaired t-test). Secondary outcomes favored the chatbot: ease of understanding (mean: 10.5/100, low difficulty), ease of use (12.24), satisfaction (77.3), confidence (82.4), and moderate clarification likelihood (50.0). No adverse events reported. The data suggests chatbots are feasible for MMS education, maintaining comprehension while boosting satisfaction/confidence, potentially reducing non-emergent inquiries. Non-significance may stem from small sample size (n=50) or self-selection bias (tech-literate participants). Interpret p>0.05 as inconclusive for superiority, but positive trends support chatbot supplements to written instructions.
Files
Steps to reproduce
This study tested whether a custom interactive chatbot improves patient comprehension of postoperative care instructions after Mohs micrographic surgery (MMS) compared to standard written instructions, conducted at the University of Miami, with IRB approval. The hypothesis was that the chatbot would enhance understanding through an engaging interface. Fifty English-speaking adult patients (≥18 years) were randomized into two groups (25 control, 25 intervention) using simple randomization via software tools, recruited during pre-surgery visits from January 2024 to June 2025. Inclusion required consent and digital comfort; exclusions included cognitive or sensory deficits. Instructions, authored by board-certified surgeons per clinical guidelines, covered wound care, activity restrictions, infection prevention, bleeding management, and emergencies, delivered either in print or via the chatbot post-surgery. Data was collected 24-48 hours later using a 10-question multiple-choice quiz (scored as percentage correct) and five 0–100 sliders for attitudes, administered via secure online forms or in-clinic, with responses anonymized in an Excel spreadsheet. Statistical analysis used an unpaired t-test (p=0.10), conducted with SPSS or R. The chatbot group scored 72.0% versus 64.4% for control, with non-significant results, while sliders showed higher satisfaction (e.g., confidence: 82.4). Trends suggest feasibility, though limited by small size and bias. To reproduce, obtain IRB approval, recruit similar patients, develop a comparable chatbot with provided instructions, deliver interventions, administer the quiz/sliders, and analyze with a t-test. Raw data is available upon request for verification.
Institutions
- University of Miami