Mr. Andre Larente reports

DIAGNOS AND ETS UNVEIL A METHOD TO ADAPT DEEP LEARNING ALGORITHMS ACCORDING TO THE DIVERSITY OF CONTEXTS

Diagnos Inc. and Ecole de Technologie Superieure (ETS), a world-renowned technical university, have released a method for adapting deep-learning algorithms according to the diversity of contexts, in diabetic retinopathy screening.

Diagnos and ETS conducted research to improve the performance of Diagnos's deep-learning algorithms in situations known to cause the degradation of learning models, particularly when the solutions are implemented in different contexts, such as in new health care facilities, when using new retinal cameras or with patient cohorts with unusual demographic characteristics. In deep learning, degradation of the learning models occurs when the data utilized to train the model differs from the data it encounter in real-world applications, which is when adaptation methods, such as those developed by Diagnos and ETS become useful.

The proposed method uses samples from the new contexts to refine the algorithms. It has proven to be a very effective solution for improving the performance of learning models. This new method brings significant improvement, so much so that the Association for Research in Vision and Ophthalmology (ARVO) elected to provide a platform to Diagnos and ETS at ARVO 2023 so that researchers, practitioners and industry stakeholders can learn about the proposed method and its major positive impacts on the field of medical imaging.

"Data is at the heart of everything we do at Diagnos. This new method of domain adaptation allows us to rapidly train our deep-learning models with diverse data sources, while maintaining and improving accuracy, making our diabetic retinopathy screening solutions more effective and versatile than ever before. We eagerly anticipate the presentation of our findings at ARVO 2023," said Riahd Kobbi, vice-president of business intelligence at Diagnos.

"At Diagnos, we always push the boundaries of innovation. Our latest research with ETS represents another step forward in our mission to improve the accuracy and speed of diabetic retinopathy screening, particularly in new and challenging environments," said Andre Larente, president of Diagnos. "As we prepare to implement our solutions into hundreds of new sites, this domain adaptation method comes at just the right time. By using deep-learning algorithms and innovative domain adaptation techniques, we are able to provide accurate and efficient diabetic retinopathy screening solutions that can be adapted to a wide range of clinical settings, thus providing huge benefits to our clients."

Diagnos also announces that Dr. Reid Maclelland has resigned from the board of directors of the corporation. The company would like to thank Dr. Maclelland for his valuable support over his term and wishes him well in his future endeavours.

About Diagnos Inc.

Diagnos is a publicly traded Canadian corporation dedicated to early detection of critical health problems based on its Flaire artificial intelligence (AI) platform. Flaire allows for quick modifying and developing of applications such as CARA (computer assisted retina analysis). CARA's AI-based image enhancement algorithms provide sharper, clearer and easier-to-analyze retinal images. CARA is a cost-effective tool for real-time screening of large volumes of patients. CARA has been cleared for commercialization by the following regulators: Health Canada, the FDA (United States Food and Drug Administration) and Saudi FDA (Saudi Arabia).

We seek Safe Harbor.