30 Seconds Of Therapy: Best Practices for Writing Clinical Prompts

START

1.  Be Specific about the condition

   1. Avoid vague prompts like “wrist pain.” The more precise you are, the more clinically useful the response.

      1.  Vague: “How do I treat wrist pain?”

      2.  Specific: “Provide physiotherapy treatment recommendations for wrist extensor tendinopathy.”

   2. Why it works: Specificity directs AI toward targeted evidence-based strategies (eccentric strengthening, manual therapy, ergonomic changes) instead of generic advice.

2. Include relevant and appropriate background information

   1. AI cannot guess the clinical picture—you need to provide it. But remember

      1. Never include patient identifiers (names, DOB, addresses, exact dates). AI tools are not secure platforms.

      2. Provide relevant anonymized details that shape the clinical context.

   2. Medical History and Background: Ex. “The patient is a 45-year-old office worker experiencing wrist pain for three months.”

      1. Previous Treatments: Ex.  “They have tried resting and using a wrist brace with limited success.”

      2. Setting and Context: Ex.“Give me physiotherapy treatment recommendations for stroke in an outpatient setting.”

      3.  Timeframe:

         1. (weak): “Describe the healing process after hip arthroplasty.”

         2.  (strong): “What can a patient typically expect during the first six weeks of healing after hip arthroplasty?”

      4. Why it works: This context makes the AI’s response clinically meaningful instead of textbook-level generic.

3. Construct prompts logically

   1. Think of your prompt like SOAP documentation:

      1. Introduction: Who is the patient (age, condition, context)?

      2. Inquiry: What do you want to know (exercises, treatment phases, risks)?

      3. Closing: Ask for extras (pitfalls, modifications, education tips).

      4. Weak:

         1. “Stroke rehab exercises?”

      5. Strong:

         1. Intro: “My patient is a 60-year-old, two weeks post-right MCA stroke, with left hemiparesis and impaired balance.”

         2. Inquiry: “What evidence-based gait training strategies can I use in early rehab?”

         3. Closing: “Include common pitfalls to avoid during therapy.”

   2. Why it works: A structured question produces a structured answer—easier to scan, apply, and adapt in practice.

STOP

1. Overly broad prompts

   1. Broad prompts waste your time and produce long, unfocused answers.

      1. Broad: “What is rehab for ACL surgery?”

      2.  Refined: “Act as a sports PT. Provide a detailed ACL post-op protocol for a 22-year-old soccer player after hamstring autograft reconstruction. Include goals, exercises, and progressions for each phase, from immediate post-op to return-to-sport.

         
         

2.  Ignoring patient privacy obligations

   1. Patient confidentiality must always come first.

      1. Wrong: “John Doe, born July 5, 1967, had a total hip replacement. What’s the best rehab plan?”

      2.  Correct: “A 57-year-old male, six weeks post-total hip replacement, reporting stiffness and mild swelling. Suggest outpatient PT interventions.”

         
         

3. Sticking to one prompt style

   1. Prompt variety improves the quality of responses.

      1. Direct: “List the early symptoms of pneumonia.”

      2. Step-by-step: “Provide a process for diagnosing pneumonia in older adults.”

      3. Summary: “Summarize pneumonia progression for patient education.”

WHY

1. AI is only as good as your input

   1. The quality of the response you get from AI depends entirely on how you frame your prompt. If you give vague or incomplete details, the answer will also be vague, broad, and less clinically useful. But if you provide clear information about the condition, relevant history, and your specific clinical question, the AI can deliver more accurate and actionable guidance. This means your role is not just to “ask a question,” but to craft it with the same precision you’d use in documentation or case presentations. Simply put, detailed input produces detailed output that supports your professional decision-making.

      
      

2. Good prompts save time and improve care

   1. A poorly written prompt often leads to long, generic answers that require extra effort to sort through and reframe. In contrast, a well-structured prompt gives you concise, relevant, and stage-appropriate information you can apply immediately. This reduces the time you spend filtering out irrelevant content and allows you to focus more on patient care. By getting straight to practical recommendations, you can enhance your workflow efficiency without compromising quality. In the end, better prompts translate into more effective care delivery and improved patient outcomes.