ChatGPT Workflow for Patient Diagnosis and Treatment

In the rapidly evolving healthcare landscape, integrating artificial intelligence (AI) tools like ChatGPT offers clinicians a unique opportunity to enhance patient care.

This blog post is a practical guide with a structured workflow that leverages AI to streamline a patient’s journey from initial presentation to the final stages of treatment and patient education.

Each section of this guide presents a specific stage in patient management – starting with the initial presentation, moving through diagnosis, exploring treatment options, considering medication choices, and addressing potential side effects.

For each stage, you will find a detailed prompt followed by a sample response illustrating how AI can assist in generating concise, relevant information.

Additionally, this guide includes critical notes for clinicians.

These notes are intended to provide context, highlight considerations specific to the healthcare setting, and offer insights into how AI can complement clinical decision-making without replacing the essential human element of care.

As a clinician, you aim to provide the best possible care for your patients. This guide shows how AI can be a valuable tool in achieving that goal, offering support in information gathering, decision-making, and patient education.

By the end of this post, you will have a comprehensive set of prompts at your disposal, which you can refer to or modify as needed in your practice.

Remember, the goal of this guide is not just to inform but also to serve as a long-term resource, enhancing the way you integrate AI into your clinical workflow for the betterment of patient care.

Configuring ChatGPT for Optimized Clinical Use

Before diving into the specific workflows, it’s crucial to tailor ChatGPT to better suit your clinical needs.

Personalizing ChatGPT enhances its relevance to your practice and improves the accuracy of its responses. This section will guide you through configuring ChatGPT, ensuring it aligns with your unique clinical environment and patient care objectives.

The customization process is where you tell ChatGPT about key aspects of your practice. This includes details such as your medical specialty, the common conditions you treat, the type of patients you see, and any specific clinical guidelines or protocols you follow.

By providing this contextual information, ChatGPT can generate more targeted and useful responses in your day-to-day clinical scenarios.

Open ChatGPT before you start. Use the settings menu to create a custom instruction.

Remember that the more precise and detailed the information you provide, the more tailored and effective the AI’s assistance will be.

You are an advanced AI assistant designed to support a private practice doctor who specializes in general patient care and has a subspecialty in dermatology. Your role is to assist in diagnosing and treating a wide range of common medical conditions while providing specialized knowledge in skin-related issues. The doctor treats patients of various ages and backgrounds, so your responses should be adaptable to diverse patient scenarios. You are familiar with the latest clinical guidelines and protocols in general medicine and dermatology, and you can provide up-to-date information and recommendations based on these standards. Your advice will be a guide, complementing the doctor’s clinical expertise and judgment in patient care.

Enter text that is similar to the example I have given you. Modify the prompt to match your expertise or situation. You will get the best results by closely aligning it with you and your practice.

Make sure you also turn on the switch for "Enable for new chats" before saving your custom instructions.

Now that we have given ChatGPT a personality, we are ready to move on to our first diagnosis.

Analyzing Symptoms: A Case Study Approach

In this section, we delve into a practical application of ChatGPT in clinical practice, using a case study to illustrate how AI can assist in the analysis of patient symptoms.

Our case involves a patient presenting with persistent hunger, frequent urination, general fatigue, and occasional tingling in the fingers and toes.

These symptoms are indicative of several potential conditions. This is a typical scenario where the AI’s analytical capabilities will help us narrow down the diagnosis.

The following sample prompt can be used as a template. It shows how to to structure inquiries for symptom analysis in a way that yields the best results.

The focus here is to use ChatGPT as a tool that will offer a preliminary analysis that can help use shape subsequent clinical steps.

Remember, while AI can provide valuable insights, it is not a substitute for professional medical judgment. The information obtained from ChatGPT should be used in conjunction with your clinical expertise and knowledge of the patient’s history and context.

Here is our sample prompt. You can replace the presentation with your own symptoms. Each part of the prompt is designed to produce a breakdown or analysis to hone in on the patients condition.

I have a patient presenting with persistent hunger, frequent urination, general fatigue, and occasional tingling in the fingers and toes. Give me a list of potential diagnoses and relevant tests or checks that should be considered with the goal of eliminating options, and identifying the specific condition my patient is experiencing. Follow this with a set of initial tests to reduce possibilities. We will do followup tests based on the results. To reduce lab test costs, include any physical or visual checks that can be used to aid diagnosis and explain the conditions that the specific physical or visual results would indicate or eliminate.

Here is a sample response. The result you get should be similar. However the results may vary based on the description of the role you provided to ChatGPT, and the model you are using (eg ChatGPT Turbo vs V4)

The symptoms your patient is presenting – persistent hunger, frequent urination, general fatigue, and occasional tingling in the fingers and toes – suggest a few potential diagnoses. I will list these along with relevant tests and physical or visual checks that could help narrow down the diagnosis.

Potential Diagnoses

  1. Diabetes Mellitus (Type 1 or Type 2)

    • Symptoms: Persistent hunger, frequent urination, and fatigue are classic symptoms.
    • Tests: Fasting blood glucose, Hemoglobin A1c, Oral glucose tolerance test.
    • Physical/Visual Checks: Look for signs of acanthosis nigricans (dark, velvety patches in body folds and creases, often in the armpits and neck), which is more common in Type 2 diabetes.
  2. Thyroid Disorders (Hypothyroidism or Hyperthyroidism)

    • Symptoms: Fatigue is a common symptom. Hyperthyroidism can cause increased appetite.
    • Tests: Thyroid-stimulating hormone (TSH) and free T4 levels.
    • Physical/Visual Checks: Check for goiter (enlarged thyroid), tremors, and signs of thyroid eye disease (in hyperthyroidism).
  3. Peripheral Neuropathy

    • Symptoms: Tingling in fingers and toes.
    • Tests: Nerve conduction studies, electromyography.
    • Physical/Visual Checks: Check for foot deformities, ulcers, and skin changes; these could be signs of neuropathy.
  4. Kidney Disease

    • Symptoms: Frequent urination can be a symptom.
    • Tests: Serum creatinine, Blood Urea Nitrogen (BUN), Urinalysis, Estimated Glomerular Filtration Rate (eGFR).
    • Physical/Visual Checks: Look for edema, especially in the lower extremities.
  5. Electrolyte Imbalance

    • Symptoms: Fatigue, tingling sensation.
    • Tests: Electrolyte panel including sodium, potassium, calcium, and magnesium levels.

Initial Tests to Reduce Possibilities

  1. Complete Blood Count (CBC): To check for anemia or signs of infection.
  2. Basic Metabolic Panel (BMP): To evaluate kidney function, electrolyte and acid/base balance, and blood glucose levels.
  3. Thyroid Function Tests: Including TSH and free T4 levels.
  4. Urinalysis: To check for signs of diabetes and kidney disease.
  5. Fasting Blood Glucose and Hemoglobin A1c: To evaluate for diabetes.
  6. Neurological Examination: If neuropathy is suspected based on initial findings.

Physical or Visual Checks

  1. Skin Examination: For acanthosis nigricans, signs of dehydration, or skin changes related to neuropathy.
  2. Thyroid Examination: Check for goiter, nodules, or signs of thyroid dysfunction.
  3. Foot Examination: For diabetic neuropathy, check for ulcers, deformities, and skin changes.
  4. Blood Pressure and Heart Rate: As both kidney disease and thyroid disorders can affect cardiovascular parameters.

By starting with these tests and observations, you can begin to narrow down the potential causes of your patient’s symptoms. The results of these initial tests will guide further testing and specialist referrals as needed. Always consider the patient’s full medical history, lifestyle, and any other symptoms when interpreting these results.

I already know that our model patient has Type 2 Diabetes. It is also important to note that the AI selected this as its primary diagnosis. However, the prompt also produced several other possibilities. These other less obvious options are often missed in a clinical setting.

At this point, the clinician would carry out a physical examination and order the tests that they feel are appropriate for the situation. If your physical examination indicates a specific condition, you can ask ChatGPT to revise the test selections as part of a normal conversation. We won’t do that in our example.

Important: Do not use your patients real name or identifying factors when you ask ChatGPT questions. It is your responsibility to maintain client privacy.

Hint 1: While you cannot absolutely rely on ChatGPT keeping your history, you should start a new conversation for every patient. The AI uses your chat history to provide answers. If you do not start a new (clean) conversation, it may inadvertantly include the previous patients symptoms in its diagnosis.

Hint 2: When you start a new conversation, ChatGPT will save your existing history in a menu at the left of the screen (or a popup menu on mobile). You can rename each of the conversations. Our recommendation is that you use the patient number to identify the conversation. DO NOT ue the patients name or other identifying factors.

Test results have arrived

We now have a set of results from the lab:

  1. Complete Blood Count (CBC)

    • Hemoglobin: 13.8 g/dL (normal range: 13.5-17.5 g/dL for males, 12.0-15.5 g/dL for females)
    • Hematocrit: 41% (normal range: 38.3-48.6% for males, 35.5-44.9% for females)
    • White Blood Cells (WBC): 6,500 /µL (normal range: 4,500-11,000 /µL)
    • Platelets: 280,000 /µL (normal range: 150,000-450,000 /µL)
  2. Basic Metabolic Panel (BMP)

    • Blood Glucose: 200 mg/dL (normal fasting range: 70-99 mg/dL)
    • Creatinine: 0.9 mg/dL (normal range: 0.6-1.2 mg/dL for males, 0.5-1.1 mg/dL for females)
    • BUN: 14 mg/dL (normal range: 7-20 mg/dL)
    • Sodium: 140 mEq/L (normal range: 135-145 mEq/L)
    • Potassium: 4.2 mEq/L (normal range: 3.5-5.0 mEq/L)
  3. Thyroid Function Tests

    • TSH: 2.5 µIU/mL (normal range: 0.4-4.0 µIU/mL)
    • Free T4: 1.2 ng/dL (normal range: 0.8-1.8 ng/dL)
  4. Urinalysis

    • Glucose: Positive
    • Ketones: Negative
    • Protein: Trace
    • Specific Gravity: 1.010 (normal range: 1.005-1.030)
    • pH: 6.0 (normal range: 4.6-8.0)
  5. Fasting Blood Glucose and Hemoglobin A1c

    • Fasting Blood Glucose: 200 mg/dL (normal fasting range: 70-99 mg/dL)
    • Hemoglobin A1c: 8.5% (normal range: <5.7%; prediabetes: 5.7-6.4%; diabetes: ≥6.5%)

I am going to assume that you have not saved your previous history. If still have a saved copy, you can re-open it to give ChatGPT the context of the previous conversation.

Here is our new prompt: (I am including the initial symptoms – on the assumption you did not save the previous chat history)

I have a patient presenting with persistent hunger, frequent urination, general fatigue, and occasional tingling in the fingers and toes. I have the following test results: [paste your results here]. What is your diagnosis along with your assumptions and/or explanation. If you are unsure, give me additional things to check.

Notice that I am asking the AI to explain the results. I am also giving it an opportunity to introduce additional tests. This will help avoid hallucinations by advising the AI that if it is not confident of the diagnosis, it can request further tests instead of making up a result on limited information.

And here is our confirmed diagnosis:

The patient’s symptoms and test results strongly suggest a diagnosis of diabetes mellitus, most likely type 2 diabetes given the absence of any indicators of type 1 diabetes such as autoimmune markers. Here is a breakdown of the relevant findings and their implications:

  1. Symptoms:

    • Persistent hunger and general fatigue are common symptoms of diabetes, as the body is unable to efficiently use glucose for energy due to insulin resistance or deficiency.
    • Frequent urination and tingling in fingers and toes can be attributed to high blood sugar levels causing diuresis and peripheral neuropathy, respectively.
  2. Complete Blood Count (CBC):

    • The CBC is within normal limits, which is helpful to rule out anemia or infection as primary causes of the patient’s symptoms.
  3. Basic Metabolic Panel (BMP):

    • Blood Glucose: The elevated fasting blood glucose level of 200 mg/dL is significantly above the normal range. This indicates hyperglycemia, a hallmark of diabetes.
    • Other parameters (Creatinine, BUN, Sodium, Potassium) are within normal ranges, indicating no immediate kidney dysfunction or electrolyte imbalance.
  4. Thyroid Function Tests:

    • TSH and Free T4 are within normal ranges, suggesting that thyroid dysfunction is not contributing to the patient’s symptoms.
  5. Urinalysis:

    • The presence of glucose in the urine (glucosuria) is consistent with diabetes, as high blood glucose levels lead to its spillage into the urine.
    • The trace amount of protein could be an early sign of diabetic nephropathy, although it’s not definitive at this stage.
    • Specific Gravity and pH are within normal limits.
  6. Fasting Blood Glucose and Hemoglobin A1c:

    • Fasting Blood Glucose: Reiterating the hyperglycemia noted in the BMP.
    • Hemoglobin A1c: At 8.5%, it is well above the diabetes threshold, indicating poor glycemic control over the past 2-3 months.

Given these findings, the diagnosis of diabetes mellitus is clear. The elevated Hemoglobin A1c suggests that this is not a new onset but a chronic condition that has been uncontrolled for some time. The next steps would include initiating a diabetes management plan, which typically involves lifestyle modifications, blood glucose monitoring, and potentially medication such as metformin. Additionally, it would be prudent to screen for common diabetes-related complications, including retinopathy, nephropathy, and cardiovascular disease.

For further confirmation and to tailor the management plan, consider:

  • Autoantibody tests (e.g., GAD, IA-2 antibodies) if type 1 diabetes is still a consideration.
  • Lipid profile, as dyslipidemia is common in diabetes and increases cardiovascular risk.
  • Referral to an ophthalmologist for a retinal exam.
  • Regular monitoring of kidney function and foot care education.

At this point we can ask ChatGPT to produce a treatment plan as a follow-on conversation (in the same chat window). It will use the context we have already given it to produce the plan.

Exploring Treatment Options: Leveraging AI Post-Diagnosis

Having conducted a thorough examination and arrived at a diagnosis, the next critical step in patient care is determining the most appropriate treatment plan.

The utility of ChatGPT here lies in its ability to provide a comprehensive overview of potential treatments, ranging from medication and lifestyle changes to more advanced therapeutic interventions.

The aim is to present you with a broad spectrum of options, enabling you to tailor the treatment plan to the specific needs and circumstances of your patient.

The next prompt will focus on the diagnosed condition. The prompt is designed to elicit a detailed response that encompassing various treatment modalities and considerations.

My patient has Type 2 Diabetes. Provide a comprehensive overview of potential treatment options. Include a range of treatments from lifestyle, medical treatments, and advanced therapeutic interventions. Provide your recommendation at the end of the output. Ask questions if you need more information to make a recommendation.

The corresponding output not only lists possible treatments but also includes key recommendations.

Treatment for Type 2 Diabetes involves a multifaceted approach, including lifestyle modifications, medical treatments, and advanced therapeutic interventions. Here’s an overview:

Lifestyle Modifications

  1. Diet:

    • Carbohydrate Management: Focus on low glycemic index foods to manage blood sugar levels.
    • Portion Control: Essential for weight management.
    • Balanced Diet: Include a variety of nutrients, reducing saturated fats, trans fats, and cholesterol.
  2. Physical Activity:

    • Regular Exercise: Aim for at least 150 minutes per week of moderate aerobic activity (like brisk walking) or 75 minutes of vigorous activity, along with muscle-strengthening activities twice a week.
    • Weight Loss: Even a modest weight loss (5-10% of body weight) can significantly improve blood sugar control.
  3. Smoking Cessation: Smoking can exacerbate diabetes complications.

  4. Regular Monitoring: Self-monitoring of blood glucose levels and regular A1c testing.

Medical Treatments

  1. Oral Medications:

    • Metformin: First-line medication, improves insulin sensitivity.
    • Sulfonylureas: Increase insulin production.
    • DPP-4 Inhibitors: Enhance insulin secretion and lower glucagon levels.
    • GLP-1 Receptor Agonists: Enhance insulin secretion, suppress glucagon secretion, and slow gastric emptying.
    • SGLT2 Inhibitors: Cause excess glucose to be excreted in the urine.
  2. Insulin Therapy:

    • May be necessary if blood glucose levels are not controlled with oral medications.
    • Various types (rapid-acting, long-acting) can be used based on the patient’s needs.
  3. Combination Therapy: Combining different classes of medications can be effective when monotherapy is insufficient.

Advanced Therapeutic Interventions

  1. Bariatric Surgery: Considered for patients with a BMI >35 kg/m² and inadequately controlled blood glucose levels, or for those with a BMI >30 kg/m² and significant comorbidities.

  2. Continuous Glucose Monitoring (CGM): For detailed blood glucose monitoring, especially useful in insulin management.

  3. Insulin Pump Therapy: Provides continuous subcutaneous insulin infusion, can be more effective than multiple daily injections for some patients.

Recommendations

Your recommendation will depend on the individual patient’s needs, preferences, medical history, and the severity of their diabetes. For most newly diagnosed Type 2 diabetes patients, starting with lifestyle modifications and metformin is standard. Regular monitoring of blood glucose and A1c levels will guide further treatment adjustments.

  • Initial Approach: Start with lifestyle modifications and metformin. Educate the patient about diabetes management, including diet, exercise, and blood glucose monitoring.
  • Progressive Treatment: If blood glucose levels are not adequately controlled, consider adding other oral medications or GLP-1 receptor agonists.
  • Advanced Management: For patients with significant hyperglycemia despite oral medications, start insulin therapy. Continuous glucose monitoring and insulin pump therapy may be considered in complex cases.
  • Regular Follow-Up: Schedule regular follow-ups to monitor the effectiveness of the treatment plan and make adjustments as needed.

It’s essential to tailor the treatment plan to the patient’s specific situation, including their cardiovascular health, kidney function, risk of hypoglycemia, and personal preferences. A team approach involving a dietitian, diabetes educator, and possibly an endocrinologist can be very beneficial.

If you have more specific information about the patient, such as their current lifestyle, BMI, comorbidities, or preferences regarding treatment, please provide it for a more tailored recommendation.

As always, the final treatment decision should be made based on a comprehensive assessment of the patient’s individual health status and in alignment with current medical guidelines and standards of care.

Documenting the Treatment Protocol

Once a treatment option has been selected, often after thorough discussions with the patient and consultation with peers, the next important step is to document the treatment protocol accurately and comprehensively.

In this step we need ChatGPT to create a clear, detailed treatment document for the patient.

The process of documenting treatment protocols is crucial for several reasons: it ensures continuity of care, aids in monitoring the patient’s progress, and provides a clear treatment outline for any other healthcare professionals involved in the patient’s care.

Utilizing ChatGPT for this task can enhance the efficiency and accuracy of this process.

The goal is to generate a response that outlines the treatment protocol in a structured format, including specifics such as medication, administration schedules, lifestyle recommendations, and any follow-up procedures or monitoring required.

I have a 48 year old male with Type 2 Diabetes. Please document the treatment protocol for: lifestyle modification with metformin and self monitored glucose monitoring. The target audience for this document is the clinician taking care of the patient.

The output provided will streamline the documentation process, ensuring that all relevant details are captured systematically and coherently. This approach not only saves time but also minimizes the risk of overlooking critical aspects of the treatment plan.

Treatment Protocol for a 48-Year-Old Male with Type 2 Diabetes

1. Lifestyle Modification

Dietary Changes:

  • Carbohydrate Management: Educate the patient on counting carbohydrates and choosing foods with a low glycemic index.
  • Meal Planning: Encourage balanced meals with controlled portions, integrating a variety of vegetables, lean proteins, and whole grains.
  • Limit Sugar and Refined Carbs: Advise to minimize intake of sugary beverages, snacks, and processed foods.
  • Healthy Fats: Incorporate sources of monounsaturated and polyunsaturated fats like nuts, avocados, and olive oil.
  • Regular Meal Timing: Encourage eating at consistent times to manage blood glucose levels.

Physical Activity:

  • Exercise Recommendation: Aim for at least 150 minutes of moderate-intensity aerobic activity per week, such as brisk walking, cycling, or swimming.
  • Strength Training: Include muscle-strengthening activities at least two days per week.
  • Activity Monitoring: Encourage the use of a pedometer or fitness tracker to monitor daily activity levels.
  • Gradual Increase: Advise the patient to gradually increase exercise intensity and duration to avoid injury.

Weight Management:

  • Goal Setting: Target a gradual weight loss of 5-10% of body weight, as this can significantly improve glycemic control.
  • Regular Weigh-Ins: Schedule regular weigh-ins during clinic visits to monitor progress.

Smoking and Alcohol:

  • Smoking Cessation: Strongly advise cessation if the patient is a smoker.
  • Alcohol Consumption: Advise moderation in alcohol intake and educate about its effects on blood sugar levels.

2. Metformin Therapy

Initial Prescription:

  • Starting Dose: Begin with a low dose of metformin, typically 500 mg once daily with evening meal to reduce gastrointestinal side effects.
  • Dose Titration: Gradually increase the dose based on glycemic control and tolerability, up to a maximum of 2000 mg per day.

Monitoring and Adjustment:

  • Efficacy Assessment: Monitor HbA1c levels at 3-month intervals to assess the effectiveness of metformin.
  • Side Effects: Instruct the patient to report any side effects, particularly gastrointestinal symptoms.
  • Renal Function: Periodically check renal function, as metformin is contraindicated in patients with significant renal impairment.

3. Self-Monitored Blood Glucose (SMBG)

Glucose Monitoring:

  • Frequency: Recommend testing fasting blood glucose daily, and 2 hours postprandial in the initial phase of treatment.
  • Target Ranges: Aim for fasting glucose levels of 80-130 mg/dL and postprandial levels <180 mg/dL.

Equipment and Technique:

  • Meter Selection: Choose a glucose meter that is easy to use and has a memory function.
  • Training: Ensure the patient receives training on the proper use of the glucose meter, including hand hygiene and finger-prick technique.

Record Keeping:

  • Glucose Diary: Encourage the patient to maintain a record of blood glucose readings, along with notes on diet, exercise, and medication doses.
  • Review of Records: Regularly review these records during clinic visits to tailor the treatment plan.

Adjustments Based on SMBG:

  • Use SMBG data to adjust lifestyle interventions and metformin dosage as needed.

Patient Education:

  • Understanding Diabetes: Provide comprehensive education about diabetes and its management.
  • Emergency Situations: Educate about the signs of hypoglycemia and hyperglycemia, and appropriate actions to take.

Follow-Up and Support

  • Regular Follow-Up: Schedule regular appointments to monitor progress, adjust treatment, and address any concerns.
  • Support Services: Refer to a diabetes educator or dietitian if needed for additional support.

Note: Adjustments to this protocol may be required based on the patient’s individual response, comorbid conditions, and preferences. Regular communication and patient feedback are key to the successful management of Type 2 Diabetes.

As with all AI-generated content, it’s important to review and customize the output to fit the specific needs of your patient and practice.

This AI-assisted documentation should be seen as a foundation upon which to build a comprehensive and personalized treatment record.

Navigating Medication Selection: A Comprehensive Approach with AI Assistance

In this section, we address the critical task of selecting medication for the patient’s treatment plan. While not always a necessary step, in our current case study, medication forms a key part of the treatment strategy.

Here, we’ll demonstrate how ChatGPT can be utilized to explore both pharmaceutical and non-pharmaceutical medication options, providing a broad perspective on available treatments.

The utilization of ChatGPT in this context aims to present an inclusive view of treatment options, encompassing traditional medications as well as alternative or adjunctive therapies.

By entering the following prompt, we will generate a response that lists potential medication choices, considering their benefits, side effects, and compatibility with the patient’s condition and overall treatment plan.

However, it’s crucial to acknowledge a significant limitation of ChatGPT in this context: its knowledge base may not be entirely up-to-date with the latest medications and medical advancements. It might be a year or two behind in terms of the latest pharmaceutical developments.

Given this limitation, the information provided by ChatGPT should be used judiciously.

I have a 48 year old male with type 2 diabetes. Provide a comprehensive list of pharmaceutical and traditional medicine options, as well as adjunctive therapies. Provide the pros and cons for each option, and possible side effects and interactions to be aware of.

It is imperative for clinicians to conduct a thorough peer review of the AI-generated suggestions and to cross-reference them with the most current medical literature and guidelines.

Comprehensive Treatment Options for Type 2 Diabetes in a 48-Year-Old Male

Pharmaceutical Options

  1. Metformin

    • Pros: First-line therapy, improves insulin sensitivity, may aid in weight loss, beneficial cardiovascular effects.
    • Cons: Gastrointestinal side effects (diarrhea, nausea), risk of lactic acidosis, particularly in renal impairment.
    • Interactions: Alcohol, iodinated contrast media; monitor renal function.
  2. Sulfonylureas (e.g., Glipizide, Glyburide)

    • Pros: Increase insulin secretion, lower cost.
    • Cons: Weight gain, risk of hypoglycemia.
    • Interactions: Alcohol, beta-blockers, and other diabetes medications.
  3. DPP-4 Inhibitors (e.g., Sitagliptin, Linagliptin)

    • Pros: Lower postprandial glucose, weight-neutral, well-tolerated.
    • Cons: Potential for joint pain, rare risk of pancreatitis.
    • Interactions: Limited, but caution with other glucose-lowering agents.
  4. GLP-1 Receptor Agonists (e.g., Liraglutide, Semaglutide)

    • Pros: Weight loss, cardiovascular benefits, less risk of hypoglycemia.
    • Cons: Gastrointestinal side effects, injection-based, cost.
    • Interactions: Slows gastric emptying, may interact with oral medications.
  5. SGLT2 Inhibitors (e.g., Empagliflozin, Canagliflozin)

    • Pros: Cardio-renal benefits, weight loss, lower blood pressure.
    • Cons: Risk of genital infections, urinary tract infections, euglycemic diabetic ketoacidosis.
    • Interactions: Diuretics, blood pressure medications.
  6. Thiazolidinediones (e.g., Pioglitazone)

    • Pros: Improves insulin sensitivity, no risk of hypoglycemia when used alone.
    • Cons: Weight gain, risk of heart failure, bone fractures.
    • Interactions: Insulin, CYP2C8 inhibitors.
  7. Insulin Therapy

    • Pros: Effective at lowering blood glucose, flexible regimens.
    • Cons: Risk of hypoglycemia, weight gain, injection required.
    • Interactions: Other diabetes medications, alcohol, beta-blockers.

Traditional Medicine Options

  1. Cinnamon

    • Pros: May improve insulin sensitivity and lower blood sugar levels.
    • Cons: Inconsistent results in studies, potential liver damage at high doses.
    • Interactions: May potentiate other diabetes medications.
  2. Fenugreek

    • Pros: May help to lower blood sugar levels.
    • Cons: Gastrointestinal side effects, limited research.
    • Interactions: May potentiate hypoglycemic effect of diabetes medications.
  3. Ginseng

    • Pros: Some evidence of blood sugar lowering effects.
    • Cons: Variable efficacy, potential interaction with other medications.
    • Interactions: Anticoagulants, MAO inhibitors, stimulants.
  4. Bitter Melon

    • Pros: Some studies suggest blood sugar lowering effects.
    • Cons: Limited research, potential liver toxicity.
    • Interactions: May potentiate hypoglycemic effects of diabetes medications.

Adjunctive Therapies

  1. Physical Activity

    • Pros: Improves insulin sensitivity, aids in weight management, cardiovascular benefits.
    • Cons: Risk of hypoglycemia in those on insulin or insulin secretagogues.
  2. Dietary Changes

    • Pros: Can improve glycemic control, cardiovascular health.
    • Cons: Requires significant lifestyle changes, possible nutritional deficiencies if not well planned.
  3. Behavioral Therapy

    • Pros: Can help with lifestyle changes, adherence to medication.
    • Cons: Requires time and commitment, access to resources.
  4. Acupuncture

    • Pros: May help with pain management and stress.
    • Cons: Limited evidence for efficacy in diabetes management, cost.
  5. Biofeedback

    • Pros: Can help manage stress and potentially blood sugar levels.
    • Cons: Requires training and equipment, limited direct evidence in diabetes management.

Conclusion

Each of these options has its unique benefits and drawbacks, and the choice of therapy should be individualized based on the patient’s medical history, preferences, and overall health goals. It’s important to monitor for potential side effects and interactions, especially when combining multiple therapies. Regular follow-up and comprehensive management involving diet

The use of ChatGPT here is intended as a supplementary tool, assisting in the initial gathering of options but not replacing the necessity for professional medical evaluation and decision-making.

This approach underscores the importance of integrating AI support with critical clinical expertise.

While ChatGPT can offer a starting point and assist in broadening the scope of potential treatments, the ultimate responsibility for medication selection lies with the clinician, ensuring the safety and well-being of the patient.

Creating Tailored Patient Education Material with ChatGPT

As we approach the final stages of our treatment workflow, an essential component is providing the patient with educational material.

This step is crucial for ensuring that patients understand their treatment, medication, and any other relevant aspects of their care. Utilizing ChatGPT for this purpose offers a significant advantage: the ability to create education material that is not only informative but also personalized to the patient’s specific situation.

In this section, we’ll demonstrate how to craft prompts for ChatGPT to generate two distinct pieces of educational content: one focusing on the treatment plan and another detailing the prescribed medication.

I have a 48 year old male with Type 2 Diabetes. Please write an education pamplet for: lifestyle modification with metformin and self minitored glucose monitoring. The target audience for this document is the patient. Include a general overview and educational content. Also include a summary of emergency situations and the actions to take. Write in Spanish.

A unique feature of using ChatGPT in this process is its capability to translate the educational material into the patient’s preferred language. This adds a layer of personalization, ensuring that language barriers do not impede the patient’s understanding of their care.

Note: We requested the information in Spanish. For clarity, I have included English output in this blog post.

Understanding Type 2 Diabetes

Type 2 Diabetes is a condition where your body does not use insulin properly, leading to elevated blood sugar levels. Managing it involves a combination of lifestyle changes, medication, and regular monitoring of your blood sugar.


Lifestyle Modifications

Diet

  • Eat Balanced Meals: Focus on fruits, vegetables, whole grains, and lean proteins.
  • Control Carbohydrates: Be mindful of the carbs in your meals and choose those with a low glycemic index.
  • Portion Size: Pay attention to the amount of food you eat to maintain a healthy weight.

Physical Activity

  • Stay Active: Aim for at least 30 minutes of moderate exercise, like brisk walking, most days of the week.
  • Build Muscle: Incorporate strength training exercises twice a week.

Healthy Habits

  • Quit Smoking: Smoking can worsen diabetes complications.
  • Limit Alcohol: Alcohol can affect your blood sugar levels.

Metformin: Your Medication Partner

Metformin helps control your blood sugar by improving your body’s response to insulin.

How to Take Metformin

  • Start Small: Begin with a low dose and gradually increase as advised.
  • With Meals: Take it with food to reduce stomach upset.
  • Regular Timing: Take it at the same time each day for the best effect.

Side Effects to Watch For

  • Common: Nausea, diarrhea, and stomach upset.
  • Serious: Severe muscle pain, difficulty breathing, or unusual sleepiness. Contact your doctor immediately if these occur.

Self-Monitored Blood Glucose (SMBG)

Keeping track of your blood sugar levels is crucial.

How to Monitor

  • Testing Frequency: Check your fasting blood sugar every morning and occasionally after meals.
  • Using the Meter: Follow the instructions on your meter for accurate readings.

Understanding Your Numbers

  • Target Range: Your doctor will provide you with target blood sugar levels.
  • Record Keeping: Write down your results to discuss with your doctor.

Emergency Situations

Low Blood Sugar (Hypoglycemia)

  • Symptoms: Shaking, sweating, dizziness, or sudden hunger.
  • Immediate Action: Eat or drink something with sugar (like fruit juice or glucose tablets).
  • Aftercare: Eat a small snack if your next meal is more than an hour away.

High Blood Sugar (Hyperglycemia)

  • Symptoms: Excessive thirst, frequent urination, or blurred vision.
  • Immediate Action: Check your blood sugar. If it’s consistently high, contact your doctor.
  • Prevention: Stick to your meal plan, medication, and exercise routine.

Diabetic Ketoacidosis (DKA)

  • Symptoms: Nausea, vomiting, abdominal pain, rapid breathing, fruity-smelling breath.
  • Action: This is a medical emergency. Seek immediate medical attention.

Living with Diabetes

Remember, managing diabetes is a team effort. Stay in regular contact with your healthcare team, attend your appointments, and keep them informed about any changes in your health. With the right approach, you can lead a healthy and active life.


Your Health, Your Journey

We are here to support you every step of the way in your diabetes management journey. Keep this pamphlet as a handy guide, and never hesitate to reach out to us with any questions or concerns.


This educational material is provided to assist in the management of your diabetes. It does not replace professional medical advice. Always consult with your healthcare provider for personal medical advice.

The example given above was very broad. You can produce more personalized educational material by describing the desired treatment plan in more detail as part of the prompt. (Eg medication doses, frequency, specific activities etc)

If you are writing in a language you do not understand, it is important to perform a reverse translation of the material back into the original language. This step is vital to verify the accuracy and appropriateness of the translation, ensuring that the meaning and nuances of the medical information are preserved.

Translate this back to English so I can check the content you have written.

It’s also crucial to discuss the varying degrees of accuracy in translations across different languages.

While ChatGPT is proficient in multiple languages, the level of precision can vary. This is particularly true for less commonly spoken languages or dialects. While Spanish, French and English produce good results, languages like Somaon, Tongan, and Maori are often incorrect or impossible for the AI to generate.

To illistrate how important this point is, I requested a patient information sheet on Type 2 Diabetes using Te Reo or Maori as the output language. It was not until I carried out the reverse translation test, that I found out it had written an information piece on breast cancer.

ChatGPT has been trained on a lot of business text. If it is unable to write what you need in your selected language, it is more likely to fall back to providing generic information from its overall knowledge base (marketing seems to be it favorite topic of choice).

You have to realize that ChatGPT is not actually a true AI. Instead it uses probability to write text based on what it has seen in the past. If it has never seen Samoan text on Diabetes, it doesn’t have the information it needs to provide an answer. Instead, it uses random probability to pick text from its huge database of Samoan language.

Because of this, clinicians should exercise extreme caution and, where possible, seek additional verification from a human translator or a language specialist, especially in cases involving complex medical information.

Summary of Key Prompts: Your ChatGPT Cheat Sheet for Clinical Practice

As we near the conclusion of our guide, it’s beneficial to consolidate the various prompts used throughout this post into a concise, easy-to-reference cheat sheet.

This summary will serve as a quick-access guide for clinicians to efficiently utilize ChatGPT in their practice.

If possible, use the same chat window for the entire conversation for a single patient. It will add context to the prompts and provide additional background information. This will result in significantly better output during each stage of the process. It also means you can ask additional questions or interact with the AI between prompts to refine the result. (EG "This patient also has hypertension.")

HIGHLY IMPORTANT: Make sure you start a new chat window when you start working on a new patient. DO NOT use the same chat window for two or more patients. The AI looks at your chat history to come up with suggestions. If you start working on a new patient in the same conversation as a previous patient, the AI may suggest the wrong treatment options (based on the symptoms of your previous patient). ALWAYS use a different chat conversation for each patient.

You can re-open old conversations using the menu built into ChatGPT. Use this to your benefit. However, make sure you have selected the correct patients notes before continuing the conversation. (Suggestion: Use the patient chart number as the reference on yor conversations – but do not use their name or other identifying information)

Below are the key prompts categorized by each stage of patient management:

  1. Configuring ChatGPT for Clinical Use:
    • Prompt to inform ChatGPT about your medical specialty, common conditions treated, patient demographics, and clinical guidelines followed.
You are an advanced AI assistant designed to support a private practice doctor who specializes in general patient care and has a subspecialty in dermatology. Your role is to assist in diagnosing and treating a wide range of common medical conditions while providing specialized knowledge in skin-related issues. The doctor treats patients of various ages and backgrounds, so your responses should be adaptable to diverse patient scenarios. You are familiar with the latest clinical guidelines and protocols in general medicine and dermatology, and you can provide up-to-date information and recommendations based on these standards. Your advice will be a guide, complementing the doctor’s clinical expertise and judgment in patient care.
  1. Analyzing Patient Symptoms:
    • Prompt for inputting patient symptoms (e.g., persistent hunger, frequent urination, general fatigue, tingling in extremities) and requesting potential diagnoses and relevant tests.
I have a patient presenting with persistent hunger, frequent urination, general fatigue, and occasional tingling in the fingers and toes. Give me a list of potential diagnoses and relevant tests or checks that should be considered with the goal of eliminating options, and identifying the specific condition my patient is experiencing. Follow this with a set of initial tests to reduce possibilities. We will do followup tests based on the results. To reduce lab test costs, include any physical or visual checks that can be used to aid diagnosis and explain the conditions that the specific physical or visual results would indicate or eliminate.
  1. Exploring Treatment Options Post-Diagnosis:
    • Prompt to generate a list of treatment options, including medication, lifestyle changes, and other interventions, for the diagnosed condition.
My patient has Type 2 Diabetes. Provide a comprehensive overview of potential treatment options. Include a range of treatments from lifestyle, medical treatments, and advanced therapeutic interventions. Provide your recommendation at the end of the output. Ask questions if you need more information to make a recommendation.
  1. Documenting the Treatment Protocol:
    • Prompt to create a detailed treatment protocol document, outlining medication dosages, schedules, lifestyle recommendations, and follow-up procedures.
I have a 48 year old male with Type 2 Diabetes. Please document the treatment protocol for: lifestyle modification with metformin and self minitored glucose monitoring. The target audience for this document is the clinician taking care of the patient.
  1. Navigating Medication Selection:
    • Prompt for listing both pharmaceutical and non-pharmaceutical medication options, considering benefits, side effects, and patient compatibility.
I have a 48 year old male with type 2 diabetes. Provide a comprehensive list of pharmaceutical and traditional medicine options, as well as adjunctive therapies. Provide the pros and cons for each option, and possible side effects and interactions to be aware of.
  1. Creating Patient Education Material:
    • Prompt to generate educational content on the treatment plan and prescribed medication, followed by translation into the patient’s language and reverse translation for accuracy check.
I have a 48 year old male with Type 2 Diabetes. Please write an education pamplet for: lifestyle modification with metformin and self minitored glucose monitoring. The target audience for this document is the patient. Include a general overview and educational content. Also include a summary of emergency situations and the actions to take.

These prompts are designed to be adapted and modified to fit the specific needs of your clinical practice and patient scenarios. They serve as a starting point for engaging with ChatGPT in a structured and effective manner, enhancing the quality of patient care through AI-assisted processes.

Conclusion: Embracing AI in Clinical Practice

In this post, we have journeyed through the innovative use of ChatGPT in various stages of patient care – from analyzing initial symptoms and exploring treatment options to documenting treatment protocols and creating tailored patient education materials.

The versatility of ChatGPT in enhancing clinical workflows, improving decision-making, and personalizing patient care has been clearly demonstrated. Each section provided practical insights into how AI can support clinicians in their daily practice, while emphasizing the importance of integrating this technology with professional medical judgment and expertise.

However, the process of manually creating prompts and interpreting ChatGPT’s outputs, although valuable, can be streamlined for even greater efficiency and accuracy.

This is where Health Connect AI, our cutting-edge SAAS product, comes into play. Designed specifically for medical use, Health Connect AI simplifies the process by allowing clinicians to input symptoms directly, automatically generating the kinds of outputs we’ve explored in this post.

Health Connect AI stands out not only for its ease of use but also for its advanced capabilities.

It has been trained to the level of a peer in the medical industry, ensuring that the responses it provides are of significantly higher relevance and accuracy compared to standard ChatGPT outputs.

Along with this, we have turned down the "randomness" factor. ChatGPT is designed for a balance between factual and fiction writing. As such, it has a high randomness factor built into its output (which is why you get a different answer each time you ask the same question) By turning down the randomness factor, we can be more confident that the output is closer to the base training set.

This additional training means that Health Connect AI is more in tune with the nuances and complexities of medical information, making it a more reliable tool for clinicians.

Moreover, we understand the paramount importance of privacy and data security in healthcare.

Health Connect AI is designed with privacy as a priority. It does not utilize entered data for training purposes, ensuring the confidentiality of patient information. Additionally, data is stored on isolated servers, providing an extra layer of security and peace of mind for both clinicians and patients.

We also offer specialized translations tools for languages that ChatGPT struggles with. For example, Maori, Samoan, Persian, and Tongan (to name a few)

In summary, while ChatGPT offers a substantial foundation for incorporating AI into healthcare, Health Connect AI takes this integration to the next level. It provides a more sophisticated, secure, and user-friendly experience, aligning with the specific needs and standards of the medical community.

As we continue to embrace AI in healthcare, tools like Health Connect AI represent the future of how technology can further enhance the quality and efficiency of patient care.

Scroll to Top