Predictive Analytics for Patient Readmission- Healthcare Data Analysis

Hospital readmissions, particularly among diabetic patients, continue to pose significant challenges in healthcare, both in terms of patient outcomes and economic impact. Recent data underscores the urgency of addressing this issue:

• Prevalence of Readmissions: Diabetes remains a leading cause of hospital readmissions. Studies indicate that adult patients with diabetes account for approximately 10% to 25% of all 30-day unplanned hospital readmissions.

• Economic Impact: The financial burden of diabetes on the U.S. healthcare system is substantial. In 2022, the total annual cost of diabetes was estimated at $412.9 billion, with $306.6 billion attributed to direct medical costs. Hospital readmissions contribute significantly to these expenses.

Addressing hospital readmissions among diabetic patients is not only crucial for improving patient care but also for alleviating the economic strain on healthcare systems. By identifying key predictors of readmission and developing predictive models, healthcare providers can implement targeted interventions to reduce readmission rates, enhance patient outcomes, and optimize resource utilization.

Dataset :

The dataset represents ten years of clinical care at 130 US hospitals and integrated delivery networks. https://www.kaggle.com/code/iabhishekofficial/prediction-on-hospital-readmission/data?select=diabetic_data.csv

Key Takeaways:

1. Specialties like Thoracic and Cardiovascular Surgery, Radiology, and Cardiology handle a high number of patients and perform more procedures per patient on average — indicating higher complexity and intensity of care.

📌 Supporting Points:

• Surgery - Thoracic tops the list with the highest average procedures (3.5) — though it has fewer cases (109), it's likely dealing with complex surgical interventions.

• Radiologists and Cardiovascular Surgeons also show high procedural involvement across large patient volumes, suggesting high diagnostic or interventional workload.

• Cardiology, with 5352 patients, is both high-volume and procedure-heavy, indicating it’s a core service line for the hospital.

2. The analysis indicates there are no substantial racial disparities in the number of lab procedures administered.

Although the average lab procedure counts vary slightly among racial groups (ranging from 40.9 to 44.1), these differences are not significant enough to suggest unequal treatment. This reflects a consistent approach to diagnostic care across patient demographics.

3. Patients with an average number of lab procedures typically had hospital stays around 4 days, showing a stable relationship between procedure count and stay duration.

However, some patients with fewer procedures still had similar lengths of stay — suggesting that not all long stays are due to high procedure frequency. These cases might represent non-procedural complexities (e.g., observation, recovery, or social factors).

🧾 Final Insight: Custom Patient Summary Analysis

I created a custom query to summarize patient data by combining medication count, lab procedures, readmission status, and race into readable profiles. Patients with the highest number of medications showed wide variation in lab procedure counts and readmission outcomes, regardless of race. This suggests that high medication usage alone isn't a clear predictor of readmission and reinforces the need for a more holistic view of patient complexity and care pathways.

Main Takeaways

• Patients typically stay less than 7 days; longer stays often involve more procedures.

• No significant racial disparities were found in lab procedure counts.

• High-procedure specialties like Thoracic Surgery and Radiology require more resources.

• Medication count alone doesn't predict readmission—patient care is multifactorial.

Call to Action:

Please feel free to connect with me on Linked In. We can have a chat or discuss more about the possible findings.