2020 Fiscal Year Research-status Report
Healthcare Risk Prediction on Data Streams Employing Cross Ensemble Deep Learning
| Project/Area Number |
20K11955
|
|---|
| Research Institution | Iwate Prefectural University |
|---|
Principal Investigator |
藤田 ハミド 岩手県立大学, ソフトウェア情報学部, 教授 (30244990)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | 知能情報学関連 / Machine Learning / Health Care System / Deep Learning |
|---|
| Outline of Annual Research Achievements |
We have used ensemble deep learning techniques by constructing Deep Neural Networks (DNNs) based on assembled CNN in architecture of two GPUs in cross
layered connection. Also, we have one GUP system, running as backup for training experiments using large scale data for comparison purpose. I could achieve good research results using zero short learning on multivariate data. Also I have trained the deep-learning architecture on dynamic data. The results was promising and therefore we publish three journal articles.
We achieved good results using . However, the accuracy for some healthcare data analytics need to be enhanced. The year 2020 was hard due to the pandemic as traveling and moving around was tight.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The experimental results provided high accuracy outcome, by testing it on the learning system, however due to the pandemic traveling restrictions, it was rather not easy to expand the results with European collaborators that I contribute with in terms of the innovative idea in Deep learning using semi-supervised learning methods. Some researchers that I collaborate are looking to test the machine learning developed model on Parkinson’s Disease: as in the joint publication: A. Butt, E. Rovini, Hamido Fujita, C. Maremmani, F. Cavallo, “Data-Driven Models for Objective Grading Improvement of Parkinson’s Disease”, Annals of Biomedical Engineering (2020) 48, p.2976-2987 https://doi.org/10.1007/s10439-020-02628-4. This is currently to be test for more accuracy on our system, also for Alzheimer.
|
| Strategy for Future Research Activity |
For the year 2021: testing data to revise the built architecture for learning. Still some multivariate health care data, are not providing high quality prediction, maybe this is related to data nature. it needs extract further features to refine the training mechanism.
Also to test multi view learning by my joint work Consensus Multi-view Clustering Model for Predicting Alzheimer’s Disease Progression" Computer Methods and Programs in Biomedicine, V.199, Feb. 2021, 105895. These new results are to be expanded and be used to detect other health issues like my work: D. Cimr, F. Studnicka, Hamido Fujita, R. Cimler, J. Slegr, "Application of mechanical trigger for unobtrusive detection of respiratory disorders from body recoil micro-movements" J.Computer Methods and Programs in Biomedicine.
|
| Causes of Carryover |
(1) 新型コロナウイルスの影響で海外出張が中止になったため
(1) Due to the pandemic the traveling for conference and meeting with other research collaborator was not possible,
(2) Some development was delayed due to traveling restrictions.
Therefore (1) and (2) need to be done in the year 2021 to construct the 2nd phase block-2, related to revise the training algorithm for better accuracy, and test it with more related data set with development team.
|
