Analyzing Determinants of Readmission

• 

  Correlation between variables

• 

  Decision Tree

Inspiration

The rising costs of healthcare and the well-being of patients are challenges that the medical community continuously strives to address. Among these challenges, hospital readmissions stand out. Not only do they represent potential inefficiencies in patient care but also they significantly increase healthcare expenses. Our inspiration stemmed from a simple question: "What factors influence a patient's likelihood of readmission?"

What it does

Our project, "Analyzing Determinants of Readmission," delves deep into patient data to uncover patterns and critical factors leading to hospital readmissions. Using supervised machine learning techniques like GLM and Decision Trees, we've created models that predict the probability of readmission and days before readmissions based on various patient parameters, from demographic details to medication specifics such as type of medicines taken.

How we built it

We began with a meticulous data cleaning process on our dataset, changing data types according to models we choose ensuring its quality and reliability. Next, we split the data into training and testing sets to ensure robust model evaluation. We then implemented two models: the GLM for treating binary dependent variable and the Decision Tree for its intuitive visual representation.

Challenges we ran into

During our analysis, the Decision Tree model produced a simplistic tree with limited number of nodes, leading to subpar predictive accuracy due to too much missing value in the original dataset.

What we learned

Beyond the technical skills of data preprocessing and model adjustment, we learned the importance of understanding the underlying data. Each data point represents a patient's journey, and our models provided a narrative of their healthcare experiences. We also realized the significance of iterative modeling – the solution isn't always the best in real-world setting, but better results are attainable with perseverance.

What's next for Analyzing Determinants of Readmission

Moving forward, we aim to incorporate more variables, perhaps from external datasets, to enhance our model's prediction. Our ultimate goal is to provide healthcare professionals with a tool that aids in reducing readmissions, ensuring patients receive the best care possible during their initial stay.

Built With

• python
• r