Machine Learning Prediction Of Readmission Within 28 Days
At AI Canberra, we believe everyone needs their daily dose of predictive analytics, so we took our data skills and machine learning methods to a rural health district in NSW battling the high rates of unexpected hospital readmissions. Although doctors and clinicians review patient information to the best of their ability before discharge, there are often little to no signs that patients are of risk of readmission. This pre-existing problem costs enormous amounts of money and resources to manage. The health district needed a solution to reduce hospital readmissions by automatically detecting potential risks and identifying groups of patients with a high rate of readmission. The solution developed by our Go Go Gadget Machine Learning team would help the hospitals to make patient-personalised interventions to reduce and manage the readmission rate.
The hospital wanted an end-to-end AI solution to classify patient records into two groups, low risk of readmission and high risk of readmission, and provide logical reasons for the classification results. Over a decade of electronic medical records were integrated with machine learning algorithms to create the final solution to meet all of the above requirements. Our team collaborated with the hospital at each step of the development process through Scrum meetings to receive feedback and revise the product to provide the most precise and customised solution.
AI Canberra has over 20 years’ experience working with data and over 10 years’ experience providing data-related solutions in the health care industry. We are uniquely positioned to ensure you get the most powerful insights out of everyday information using health-focused intelligent agents. It’s just what we do. Our experience demonstrating the ROI of AI in rural health districts makes us the perfect partner to transform your health initiative.
Data is the new oil, with AI as the pistons in the engine that power your everyday business decisions. An AI powered patient risk machine learning model was developed to analyse features of patient demographic, health condition history, and treatment results during admission. Upon analysing the data, the model could predict which patients had a potential high readmission risk, and an explanation for the predicted severity. We took the model to the clinicians, supporting real-time decisions on SQL Server on premise with Microsoft SQL Server R Services. The final results were made to look magnificent, presented in Microsoft Power BI wrapped in beautiful LiveTiles SharePoint pages.
It took one month to understand the client’s needs and set up the project goal; one month to collect sufficient data and perform data engineering; two months to build predictive models and select the best performing product; one month to refine the final model; and one month to deliver the results as a product.
When test data at a daily base was applied to our trained model, the model had an overall seventy-percent accuracy in targeting patients who had an unexpected readmission within 28 days of their initial admission.
This solution proved that machine learning methods can demonstrate the immense value of data, to automatically detect potential risks and identify groups of patients that have a high risk of readmission. Hospitals can use AI powered decision making tool to assist doctors and clinicians to make real time decisions to prevent readmission after discharge.
As each decision is made, the savings add-up to several thousand dollars per readmission, creating funding to solve the next problem. Clinicians were now more engaged with factors that cause readmissions, and continued to use their experience to refine the model to greater accuracy. This project demonstrated an Office 365 based architecture for predictive analytics and machine learning methods to analyse electronic medical records to reduce the readmission rate of high risk patients. AI is no longer a secret to success, but now an obtainable solution to transform yesterday’s problems into a brighter tomorrow.