81% Patient Comprehension: What We Learned

The Health Literacy Crisis

In the United States, nearly 90 million adults have low health literacy—meaning they struggle to understand medical information, follow treatment plans, and make informed decisions about their health. This isn't an education problem; it's a communication problem. Doctors write in medical jargon. Patients read discharge summaries filled with phrases like "bidirectional atrioventricular nodal reentrant tachycardia" and nod along, confused.

The consequences are staggering: low health literacy is associated with higher hospital readmission rates, worse medication adherence, more preventable complications, and lower quality of life. Studies show that patients who don't understand their diagnosis are 40% less likely to take prescribed medications correctly and 2x more likely to have preventable ER visits.

When we started Synthure, our hypothesis was simple: what if we could translate doctor's notes into language patients actually understand? Not just dumbed-down, but clinically accurate and patient-centered. We tested this with 200 beta patients and the results exceeded our expectations.

Study Design: 200 Patients, Pre/Post Evaluation

We conducted a randomized controlled trial across three hospital systems (cardiology, psychiatry, internal medicine) from October 2025 to January 2026. Study participants:

• 200 patients diagnosed with a new condition (heart failure, depression, diabetes, hypertension)
• Randomly assigned to control (standard discharge notes) or intervention (Synthure AI translation)
• Assessed at discharge and 4 weeks post-discharge

Primary Outcome: Patient Comprehension

We measured comprehension using the PEMAT (Patient Education Materials Assessment Tool), a validated instrument for assessing understandability and actionability of health information. Patients were asked to:

1. Recall their diagnosis in their own words
2. Explain the primary treatment or medication
3. Identify warning signs that require urgent care
4. Describe lifestyle changes recommended

Responses were scored 0-100 by clinical raters using a standardized rubric. A score of 70+ indicates "clinically acceptable" understanding.

Secondary Outcomes: Real-World Impact

Comprehension is important, but outcomes are what matter. We tracked patients for 4 weeks post-discharge and measured:

The Medication Adherence Story

One of our most striking findings was the jump in medication adherence. In the baseline group, 39% of patients reported skipping doses or stopping medications early—usually because they didn't understand why they needed them or feared side effects they'd heard about.

When Synthure translated a cardiologist's note on "ACE inhibitors for systolic dysfunction" into "This blood pressure medicine helps your heart pump more efficiently. Common side effects include a mild dry cough, which is harmless and goes away in most patients," adherence jumped. Patients understood not just what to take, but why and what to expect.

ER Visits and Hospital Readmission

The 52% reduction in preventable ER visits is particularly significant. We defined "preventable" as visits for issues that could have been managed at home with proper understanding (e.g., a diabetic patient going to the ER for mild hyperglycemia that could've been addressed by adjusting diet).

The 23% reduction in readmission is clinically meaningful and economically significant. Hospital readmission is expensive (~$15K per admission) and often preventable with proper post-discharge education and monitoring. Our data suggests plain language translation helps patients stay out of the hospital.

Subgroup Analysis: Who Benefits Most?

We stratified results by age, education level, and health literacy at baseline. Interestingly, the benefit was strongest in vulnerable populations:

This is crucial: the patients who struggle most with standard medical information benefit most from AI translation. Elderly patients, those with limited education, and those with low baseline health literacy saw 70+ percentage-point improvements. Even college-educated patients benefited, but the relative gains were smaller.

Qualitative Feedback: What Patients Said

Beyond numbers, we gathered qualitative feedback. Representative quotes from Synthure patients:

Cost-Effectiveness Analysis

From a healthcare economics perspective, Synthure's impact is substantial. The average cost of one hospital readmission is $15,000. A 23% reduction in readmissions across 200 patients translates to 9 prevented readmissions, or $135,000 in avoided costs over 30 days.

The cost to deploy Synthure per patient is approximately $5 (inference on our GPU infrastructure). This implies a return-on-investment of 27:1 in readmission cost reduction alone—not accounting for the value of prevented ER visits, improved quality of life, or avoided medication errors.

Limitations and Future Work

Our study has limitations worth noting:

• Short follow-up: We tracked patients for 4 weeks. Long-term impact (6-12 months) remains unknown.
• Hospital setting: Our cohort was mostly urban, insured patients in academic medical centers. Results may not generalize to rural or underinsured populations.
• Disease diversity: We focused on three conditions. Impact on rare diseases or complex multimorbidity is unexplored.
• Hawthorne effect: Being in a study may have biased patients toward better adherence. Real-world impact could be smaller.

Next steps: We're planning a larger (1,000-patient) multi-site trial with 12-month follow-up, expanded to include primary care and chronic disease management. We're also exploring integration with electronic health records to make Synthure a seamless part of clinical workflow.