2023 Fiscal Year Annual Research Report
Development of a novel method for prediction using artificial image and image identification
Project/Area Number |
22K21186
|
Research Institution | Fujita Health University |
Principal Investigator |
HE YUPENG 藤田医科大学, 医学部, 助教 (00953267)
|
Project Period (FY) |
2022-08-31 – 2024-03-31
|
Keywords | Artificial image / Model performance / Prediction / Epidemiology / Feature sequence |
Outline of Annual Research Achievements |
This study aims to develop an novel approach to enhance model precision using artificial images. We are sincerely grateful to KAKENHI for funding this research. In the first fiscal year, we developed an algorithm to transform tabular data into pixel values, and algorithm to construct artificial images by sequencing pixels. We discovered that the sequence of features (pixels) can affect model accuracy when using AI-based machine learning techniques. These findings were presented at a small-scale conference. In the second fiscal year, we constructed artificial images using the aforementioned algorithm, and applied this approach to construct a classifier for schizophrenia using online survey data. The classifiers created using our novel approach achieved an average accuracy of 0.88 (AUC: area under the receiver operating characteristic curve). A small number of classifiers even reached an accuracy higher than 0.93, confirming the existence of an optimal artificial image and an optimal model. These findings were presented at the 34th annual scientific meeting of the Japan Epidemiology Association, and the research article was published in JAMIA Open (2024, volume7, issue 1). The concept of this approach has been submitted to the Japan Patent Office for patent application.
|