Abdulaziz Al-Homaid. Image: HBKU
Qatari researcher Abdulaziz Al-Homaid is using AI to discover more about population-specific risk factors for diseases such as diabetes.
Software engineer Abdulaziz Al-Homaid is researching state-of-the-art deep learning in order to find applications for this technology in areas such as healthcare.
With an MSc in data science and engineering from Hamad Bin Khalifa University (HBKU) in Doha, Qatar, he is now based at the university’s Qatar Computing Research Institute (QCRI).
Previously, during a fellowship with the United Nations Office for the Coordination of Humanitarian Affairs, Al-Homaid worked on improving access to education data across response organisations.
Now, his research is exploring how AI and machine learning might plug into healthcare, with a focus on population-specific needs. For example, one application of his research could better support the management of diabetes when patients are fasting during Ramadan.
‘The hope is that this research will translate in the MENA region due to similar characteristics and lineage’
– ABDULAZIZ AL-HOMAID
What inspired you to become a researcher?
While I was working as a web developer for a local company, I was fairly interested in reading about pattern recognition systems and listening to podcasts on how AI works. I was lucky enough to find out about the data science and engineering programme at HBKU. Pursuing the programme led me to practically learn more about machine learning specifics and see countless opportunities of its applications and its ability to automate processes on a big scale.
I am particularly inspired by AI research that positively impacts human work and connectivity, which, in turn, helps us understand ourselves better.
What research are you currently working on?
As a software engineer in QCRI, my goal is to enable computational health research by combining multimodal data and developing technological solutions to support aspects of precision health. This includes dealing with data related to lifestyle and behavioural data retrieved from internet of things devices – wearables to track activity and vitals such as heart rate and sleep, and weight scales – as well as environmental and cultural factors.
As a team, we conduct our work with the local Qatar health stakeholders (hospital system, policy makers, public health) in order to improve the overall health and outcomes in the population. The hope is that this will translate in the MENA region due to similar characteristics and lineage.
We are trying to answer specific questions such as: could European and other population studies be translated to the Qatari population? What are the local population-specific risk factors for a particular disease like diabetes? What role does Qatari lifestyle including activity, dietary habits, sleep behaviour play on risk and progression of disease? Can we tailor interventions for individual behaviour modification (eg in childhood obesity) using artificial intelligence techniques and IoT devices, and what impact does that have on outcomes?
In your opinion, why is your research important?
Successful AI models applied to the e-health domain complement the staff expertise in the healthcare sector. For instance, a physician may crunch the numbers of thousands of health records using a deployed AI model to provide statistically accurate correlations between patients’ lifestyle and physical activity. With this gained information, the physician is enabled to make clinically relevant decisions tailored to each specific individual.
Furthermore, the physician could use the technology developed to monitor the treatment of the individual and provide recommendations digitally without making physical appointments.
What commercial applications do you foresee for your research?
One application could be the technology to predict future hypoglycaemic (low blood sugar) episodes for people with diabetes, tailored specifically for the local region lifestyle.
Receiving insights on the glucose level relieves the stress of having hypoglycaemia and helps the individual to better prepare during the day, especially during fasting seasons like Ramadan.
What are some of the biggest challenges you face as an AI researcher?
As a researcher who deals with patients’ health data on a daily basis, I am required to use de-identified data with no known sharing or use restrictions. Thus, proper care is required at all times while using the data.
This translates to the challenging requirement of the knowledge to develop privacy-preserving technologies at all stages of a system. It gets harder when multiple developers are working on the same project as access to the data is strictly limited.
Are there any common misconceptions about AI research?
A common public misconception is that AI applications may take jobs away from people. Artificial intelligence is extremely far away from achieving human general intelligence. While AI may surpass humans in extremely specific tasks such as in the strategy game Go, the same AI cannot solve simple brain teasers.
Each job is made up of a collection of tasks, many of which are not easily automated. On the contrary, AI will help complement these jobs and make some dull and repetitive tasks easier to do. Consider matching jobs for jobseekers – an AI could be used to find available vacancies based on the user’s résumé.
What are some of the areas of research you would like to see tackled in the years ahead?
To understand why a machine learning model predicted some output, one would require computational knowledge and apply different interpretation strategies. While there exists some work on machine learning interpretability, the research area still lacks the ability to explain a model’s predictions to experts in other fields or at the level of an institution.
This area of research would require multidisciplinary intervention including, but not limited to, domain experts, human-computer interaction professionals and data scientists.
Want stories like this and more direct to your inbox? Sign up for Tech Trends, Silicon Republic’s weekly digest of need-to-know tech news.
Are you a researcher with an interesting project to share? Let us know by emailing editorial@siliconrepublic.com with the subject line ‘Science Uncovered’.
