MASTER Assignment
Mortality prediction after hip fracture surgery. Could Machine Learning add value?
Type: Master CS
Duration: TBD
Student: unassigned
If you are interested, please contact:
Introduction:
Hip fractures are a significant health care problem in the elderly, affecting 1.5 million people per year worldwide. Over 90% of hip fracture patients are older than 65-year-old and have preexisting medical comorbidities. Both factors have an important influence in its prognosis and treatment.
The consequences of a hip fracture can be serious; one-third of the patients die within the first year postoperative. The mortality rate is highest in the early postoperative period, reaching up to 13.3% within the first 30 days after surgery. Identification of patients at high risk of early mortality could improve the quality of care.
The Almelo Hip Fracture Score (AHFS) is knowledge based model to assess the risk factor mortality after surgery. This model is based on quantitative measures from 850 patients of which the data is collected in structured format within the medical record. Potential relevant clinical information in non-structured format (eg. cardiology or radiology reports) within the medical record is not taken in account.
The goal of the assignment is to extend and hopefully improve the AFHS by incorporating textual patient information from other specialist domains by using Machine Learning.
Tasks:
- Gather and organize relevant data.
- Analyze the quality of the data.
- Visualize the data.
- Transform textual information to a useful feature space.
- Integrate the textual information in the AHFS model using Machine Learning.
- Compare with Machine Learning models which are learned from scratch, i.e. not using AHFS.