Artificial Image Objects (AIO) for Disease Diagnosis and Risk Management

To identify a specific biological trait from a large volume of genetic and other biological omics data, UTHealth researchers developed a general method to transform the biological data into artificial image objects (AIO). The artificial intelligence (AI)-assisted analysis of AIOs based on image classification can quickly determine the presence of different biological traits. The AIO technology can be broadly applied to disease prediction, diagnosis, treatment, and prevention.

Background

Over the last 50 years, researchers have discovered that various genetic variations, gene expression patterns and levels, and varied protein functions can have different impacts on human diseases. However, it remains a great challenge to identify specific variants that are responsible for different diseases from a large dataset. To address this problem, the UTHealth researcher developed a technology that is capable of transforming a large volume of genetic and other biomarker data into actionable information for diagnosis, treatment, and prevention of disease.

Benefits/Technology Advantages

• The AIO technology is capable of processing large volumes of datasets via a simple machine-learning algorithm, which saves time and cost for genetic diagnosis.
• The AIO technology can also protect the private omics data of patients by transforming the data into digital images.
• The AIO technology can deal with genetic datasets and other biomarker omics datasets, including protein data, epigenomic data, microbiota data, proteome data, etc.

Potential Applications

• Hospitals or genome data analysis companies can upgrade their bioinformatic system with an incorporated AIO technology. This would allow patients to receive their diagnostic test results earlier and save time for disease treatment and prevention.
• Healthcare institutions can also use this technology to discover the underlined relationship among different genetic variations, expression levels, and biological phenotypes for drug discovery and development.
• This technology can also be applied to personalized drug screening and prediction of clinical outcomes of therapeutic treatments, eventually for individualized selection of the best therapy.

Intellectual Property Status

U.S. and international patents pending, U.S. Publication No: US 2020/0381083 A1

Available for licensing.

About the Inventor

Xiangning (Sam) Chen, Ph.D.

Professor of McWilliams School of Biomedical Informatics (SBMI) at UTHealth

Dr. Chen’s research interest is to model disease risk, facilitate early and accurate diagnosis, and provide information for targeted and personalized intervention and treatment.

UTHealth Ref. No: 2023-0025