Healthcare organizations are turning to data to gain insights to make accurate predictions on patient outcomes. A recent customer wanted to use AI/ML to identify patient wounds that won’t heal. These insights are valuable for healthcare organizations and help doctors understand the key drivers that help wounds heal and provide better medical care and treatment to their patients.
Solution
To solve this problem, we created a classifier to identify if the wound will heal or not. Given a new patient’s health history and lifestyle, our model can accurately assess if a wound will heal, giving medical professionals information to suggest treatments to the patient.
Process
1. Extracting Data
The dataset was collected and stored in Google’s BigQuery. It was rich with useful information but was spread out in a variety of different tables, which required complex JOIN clauses to bring the information into one centralized dataset.
2. Data Exploration
Although a great deal of research was required to understand the features and the medical terms, medical experts guided us on the various nuances in the wound healing process.
3. Feature Engineering
With the help of subject matter experts, we were able to identify and create new useful features for model predictions.
4. Model Training
We trained the model on different classification algorithms and evaluated its performance on a test set.
Here are the results of our best model:
| Evaluation Metrics | Percentages |
| Accuracy | 90.97% |
| Precision | 91.90% |
| Recall | 89.96% |
| F1 Score | 90.92% |
Future Work
Depending on our customer’s interest, there are possibilities to re-train the model to reduce the number of features in the dataset while still holding its predictive accuracy. For online and real-time predictions, the model can be hosted on Google Cloud and integrated into the customer’s CRM platform.
If you have any questions or have examples you’d like to share, we’d love to hear what tools you find useful. Drop us a line at [email protected] or tweet us @springmlinc.
About SpringML
SpringML delivers data-driven digital transformation outcomes with an experimentation and design thinking mindset. We provide Google Cloud consulting and implementation services and industry-specific analytics solutions that deliver high-impact business value from data. SpringML is a Google Cloud partner with capabilities to plan, assess, deploy, and manage data-driven engagements. We have been awarded Google Cloud specialization based on our expertise and customer portfolio for Data Analytics, Machine Learning, and Marketing Analytics.
