Different types of data are typically collected during AD studies — phenotype, genetic and imaging, among others. Gaining insight from the available data requires analytical skills and competencies in many fields. The underlying complexity of the disease makes it impossible for a single research group to achieve on its own scientifically substantial results in finding the causes of the disease or a cure for it. Because of that, at GATE we work together with colleagues from leading universities in the EU, US and UK on several problems, related to AD. The main research topics in the Digital Health domain include:

• Omics-type of analytics: combining different types of biomedical data in order to gain insight into particular diseases. One of our main projects includes the development of ML-based methods for the analysis of genomic data. The goal is to create a fine-tuned classification of Alzheimer’s disease patients and to predict the speed of cognitive decline.
• Risk calculator for estimating the risk of AD that will be used for pre-screening in order to avoid unnecessary costly and invasive medical procedures.
• Analysis of Next Generation Sequencing data: development of analytical pipelines, improvement and evaluation of existing methods for DNA-Seq and RNA-seq analysis. Such methods include structural variant detection, differential expression and splice variant analysis.
• Synthetic data: development and evaluation of methods for the generation of synthetic data with specific predefined properties that mimic real-world biomedical data. An important use of synthetic data is to enhance real-world data and to improve the prediction capabilities of the analytical methods that are applied to them. Synthetic data is also used for the evaluation of analytical techniques and their properties as well as selecting the optimal methods in different scenarios.
• Missing data imputation in clinical or epidemiological research: ignoring observations with missing data leads to loss of precision or introduces bias. Our theoretical research includes analysis of existing imputation methods as well as the development of novel methods for imputation of missing data.
• Information systems related to the Digital Health domain: an example is the web-based CogniSoft system that provides digital diagnosis and rehabilitation of cognitive impairment in parents with cognitive diseases.
• Some of our other projects include a pilot project on developing and applying ML methods to single-cell sequencing data, modelling of COVID-19 data, analysis of data from IoT devices and mobile apps and text analytics of medical records.

Our pilot project in the field of Digital Health is related to the use of artificial intelligence as a tool for the diagnosis and treatment of cognitive diseases with a focus on Alzheimer’s disease.

The main goal of the project is to develop methods, algorithms and tools that support risk prediction, diagnosis and treatment of cognitive diseases and in particular Alzheimer’s disease. The expected results from a research point of view are related to finding new classifications of patients with Alzheimer’s disease, developing models for predicting a cognitive decline, identifying structural variants of DNA in patients with Alzheimer’s disease, defining the necessary sample sizes in the planning of RNA research, development of optimal methods for finding genes that could be associated with Alzheimer’s disease, as well as methods for filling in missing data. Practitioners are expected to develop tools to facilitate the diagnosis, monitoring and treatment of patients with Alzheimer’s disease or people predisposed to it.
The developed methods, algorithms and tools can subsequently be integrated into an omix analytical model.

The project is implemented through successful cooperation with leading universities at the international level such as Yale, Oxford, Stanford and Chalmers, as well as with the support of medical professionals in Bulgaria.

“Opportunities and Prospects in Digital Health” session, GATE Big Data and Artificial Intelligence Forum, December 10th, 2020

The goal of the “Opportunities and Prospects in Digital Health” session of the GATE Big Data and Artificial Intelligence Forum was to discuss various elements of the process of utilizing, analyzing and gaining insight from Big Data in the Digital Health domain. The emphasis was on collaboration between different stakeholders — biomedical researchers, hospitals and medical doctors, data scientists, biostatisticians and bioinformaticians, pharmaceutical companies, etc. Consequently, the panellists were representatives of different fields within the Digital Health domain.

The keynote speaker of the session was Prof. Dr Milan Petkovic, Vice president of BDVA and the Data Science department head at Philips. Having a unique insight as part of the industry, academia and the ecosystem, his presentation underlined the current state and the perspectives in the Digital Health domain. He emphasized the trends in healthcare that are the main driver for innovation; the new reality in digitalization and telemedicine; the technical changes, such as the growth of health data; also the collaboration between different entities by utilizing not only analytical methods but also specific domain knowledge. In addition, Prof. Petkovic showed several specific use cases, including AI-enabled solutions and described several EU initiatives in the field.

The session moderator Dr Dean Palejev, Digital Health Research Lead at GATE, presented some of the institute’s main projects in the domain. These include developing AI and ML-based methods for analyzing Next Generation Sequencing data and are implemented in collaboration with researchers from the US and Sweden.  The development of NGS over the past decade made this type of data one of the richest sources of genomic data and, together with the advancements in the Big Data, AI and ML analytical methodology opened vast research possibilities in the field.

The vice-chair of the Board of Digital Health and Innovations Cluster Bulgaria, Dr Rosen Dimitrov presented his organization and its members representing various companies and organizations in the field. He emphasized their vision for creating a data-driven ecosystem in the Digital Health domain in the country. He also described the services provided by his organization including access to information, mentoring and support and importantly, providing a platform for collaboration between different types of stakeholders in the Digital Health ecosystem.

The next panellist was Dr Fredrik Johansson, Assistant Professor at the Chalmers University of Technology who outlined some advanced initiatives at his university. One such example is the Chalmers AI Research Centre (CHAIR) that also includes external collaborators from the industry and the public sector and Dr Johansson described in more detail their collaboration with Sahlgrenska University Hospital. He also mentioned his work on predicting the cognitive decline of Alzheimer’s patients, part of which is a collaboration with GATE.

As a medical practitioner Dr Todor Kunchev, a neurologist at University hospital Sofiamed shared his thoughts on the opportunities of digitalization of dementia management prompted by the increased number of cases due to increased life expectancy. He presented an example of such a system, the CogniTwin project on which he is partnering with GATE. The project was designed to create a digital twin model for the exploration of behavioural changes in patients with proven cognitive disorders and its implementation utilizes modern Big Data and AI technologies.

At the end of the session, the panellists discussed the important recent advancements in the field from their point of view, as well as their predictions for the next few years. As expected it became apparent that different technologies and trends are important for the different participants in the ecosystem, and that the synergy of rapidly developed technologies is the leading force of innovation in the Digital Health domain.