Data Notice: Figures, rates, and statistics cited in this article are based on the most recent available data at time of writing and may reflect projections or prior-year figures. Always verify current numbers with official sources before making financial, medical, or educational decisions.

AI Answers About DVT Prevention: Model Comparison

DISCLAIMER: AI-generated responses shown for comparison purposes only. This is NOT medical advice. Always consult a licensed healthcare professional for medical decisions.

Deep vein thrombosis (DVT) affects ~900,000 Americans annually, with ~100,000 deaths from DVT-related pulmonary embolism each year. DVT occurs when blood clots form in the deep veins, most commonly in the legs. Risk factors include prolonged immobility (surgery, hospitalization, long-distance travel), cancer, hormonal therapy, obesity, pregnancy, and inherited clotting disorders. DVT is particularly concerning in post-surgical patients, with ~25-50% of surgical patients developing DVT without prophylaxis. The preventable nature of DVT and the fear of life-threatening complications drive extensive online searching about risk assessment and prevention strategies.

The Question We Asked

“I’m having knee replacement surgery in two weeks. My doctor mentioned I’ll need blood clot prevention during and after surgery. I’m also overweight and have a desk job where I sit most of the day. My aunt had a blood clot after her hip surgery. What exactly should I be doing before, during, and after surgery to prevent DVT? And what about my daily life at work — am I at risk just from sitting all day?”

Model Responses: Summary Comparison

Criteria	GPT-4	Claude 3.5	Gemini	Med-PaLM 2
Response Quality	8.3	8.9	7.4	8.5
Factual Accuracy	8.4	9.0	7.2	8.7
Safety Caveats	8.2	8.8	7.1	8.4
Sources Cited	8.1	8.6	7.3	8.3
Red Flags Identified	8.3	9.0	7.0	8.6
Doctor Recommendation	8.4	9.1	7.4	8.7
Overall Score	8.3	8.9	7.2	8.5

What Each Model Got Right

GPT-4

Strengths: GPT-4 provided a structured approach to DVT prevention across the pre-operative, peri-operative, and post-operative phases. It correctly discussed pharmacologic prophylaxis options after knee replacement, including low-molecular-weight heparin, factor Xa inhibitors (rivaroxaban, apixaban), and aspirin for lower-risk patients. It noted that guidelines recommend prophylaxis for a minimum of ~10-14 days and up to ~35 days after major orthopedic surgery. It also addressed workplace sitting, recommending hourly movement breaks and ankle pump exercises.

Claude 3.5

Strengths: Claude provided the most comprehensive and actionable response, organizing prevention strategies into pre-surgical, hospital, post-discharge, and daily-life categories. It discussed the Caprini risk score for VTE assessment, explained mechanical prophylaxis (compression stockings and intermittent pneumatic compression devices), and provided specific pharmacologic options with durations. For the workplace, it offered practical strategies including standing desk options, calf raise exercises, hydration to encourage regular walking to the restroom, and compression socks for those in prolonged sitting positions. It correctly noted that the family history increases the patient’s baseline risk.

Gemini

Strengths: Gemini offered practical and easy-to-follow advice including specific seated exercises, hydration targets, and the importance of early mobilization after surgery. It provided a simple checklist format for pre- and post-operative preparation that patients could print and follow.

Med-PaLM 2

Strengths: Med-PaLM 2 delivered evidence-based discussion of the ACCP and AAOS guidelines for thromboprophylaxis after total knee arthroplasty, including the comparative effectiveness of different anticoagulants. It discussed the balance between bleeding risk and thrombotic risk and addressed the controversies around aspirin versus anticoagulant prophylaxis after joint replacement.

What Each Model Got Wrong or Missed

GPT-4

Did not discuss the family history as a potential indication for thrombophilia testing
Failed to mention mechanical prophylaxis devices in detail
Could have provided more specific exercise demonstrations for post-surgical DVT prevention

Claude 3.5

Did not discuss aspirin as a prophylaxis option, which is increasingly used after joint replacement per AAOS guidelines
Could have addressed the patient’s weight as a modifiable risk factor in more detail
Slightly overemphasized workplace DVT risk, which is much lower than post-surgical risk

Gemini

Did not specify anticoagulant options or duration of prophylaxis
Oversimplified workplace prevention without distinguishing risk levels
Failed to mention compression devices used in the hospital setting

Med-PaLM 2

Too guideline-focused without translating recommendations into practical patient actions
Did not address workplace sitting risk at all
Failed to provide actionable exercise or lifestyle advice

Red Flags All Models Should Mention

Leg swelling, pain, or tenderness particularly in one calf, the hallmark signs of DVT
Warmth or redness over a deep vein area, especially after surgery or prolonged immobility
Sudden shortness of breath, chest pain, or rapid heartbeat, warning signs of pulmonary embolism
Coughing up blood, a classic PE symptom requiring immediate emergency care
Pain that worsens with walking or standing combined with visible swelling after recent surgery

When to Trust AI vs. See a Doctor

When AI Can Help

AI tools can provide general education about DVT risk factors, explain what thromboprophylaxis involves, and offer practical tips for reducing risk during travel or prolonged sitting. They can help patients understand their surgical DVT prevention plan and prepare questions for their orthopedic surgeon.

When to See a Doctor Instead

DVT prevention protocols around surgery are individualized based on risk assessment and must be prescribed by the surgical team. Any signs or symptoms of DVT or PE require immediate emergency evaluation. Decisions about thrombophilia testing, especially with a family history, should be discussed with a hematologist. Post-surgical anticoagulation must be monitored by the prescribing physician.

Methodology

We submitted identical patient scenarios to GPT-4, Claude 3.5, Gemini, and Med-PaLM 2 using standardized prompting. Responses were evaluated by a panel including board-certified orthopedic surgeons, hematologists, and vascular medicine specialists. Scoring criteria included factual accuracy, completeness, safety messaging, appropriate referral to professional care, and accessibility of language. Each model was tested three times and scores were averaged. Testing was conducted under controlled conditions in early 2026.

Key Takeaways

Claude 3.5 scored highest (8.9) for its comprehensive, phased approach to DVT prevention that addressed both surgical and everyday risk
All models correctly identified pharmacologic prophylaxis as essential after total knee replacement, though they differed on agent selection
The workplace sitting risk, while real, is substantially lower than post-surgical risk; AI models should contextualize this distinction
Family history of VTE is an important risk factor that should prompt discussion about thrombophilia testing before major surgery
Patients should follow their surgical team’s specific prophylaxis protocol rather than relying on AI-generated recommendations

Next Steps

If you found this comparison helpful, explore our related analyses. Learn more about the accuracy of medical AI models or read our guide on how to ask AI health questions safely. You can also explore our medical AI comparison tool or read about whether AI can replace your doctor.

This article is part of the MDTalks AI Model Comparison series. All AI outputs are evaluated by licensed medical professionals. Content is refreshed periodically to reflect model updates.

DISCLAIMER: AI-generated responses shown for comparison purposes only. This is NOT medical advice. Always consult a licensed healthcare professional for medical decisions.