Empowering decision support in healthcare with AI
Machine Learning has increasingly become an important tool for existing and new software systems in the healthcare industry [1]. Here, we showcase an End-to-End Critical Care Management System (CCMS) that focuses on providing practitioners a platform to predict the probability of death of the patie...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2020
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/144799 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Machine Learning has increasingly become an important tool for existing and new software systems in the healthcare industry [1]. Here, we showcase an End-to-End Critical Care Management System (CCMS) that focuses on providing practitioners a platform to predict the probability of death of the patient in ICU based on observed mortality rate from historical patient data using a novel federated learning boosting technique called LoAdaBoost. The prediction of each patient’s mortality rate is learned from the representation space of patient information and the prescriptions taken by them. During the process of training the model, LoAdaBoost would be incorporated into traditional federated learning technique to increase the overall model’s efficiency and performance. The critical care data are further split into IID and non-IID data distribution to investigate and observe the performance level and eventually prove that using the LoAdaBoost method could achieve higher performance with lower computational complexity than the classical FedAvg and later be chosen as the model to train CCMS data information. |
|---|