Enhancing Population Health Screening Through Artificial Intelligence

As per the estimates of the World Health Organization, about 41 million people die each year globally because of non-communicable diseases. A sizable portion of people do not know that they have potentially fatal illnesses or serious underlying health conditions. Post-COVID-19, the world has also realized the high human and economic cost of infectious diseases. Therefore, early detection and identification are key for timely treatment, curtailing the growth of the disease and lowering morbidity and mortality.

The challenge, however, is that most often patients do not have access to physicians in low- and middle-income countries. Inadequate access to diagnostics tools is another major barrier to screening in under-resourced populations. Additionally, many diagnostic platforms are expensive, invasive, or require trained healthcare workers to operate, making them infeasible for deployment in resource-limited settings.

Artificial intelligence to the rescue

Population health screening through artificial intelligence (AI) is a promising way to address these challenges. In a coordinated programme, population screening entails providing a test to every member of an eligible group.

AI offers a way to improve the efficiency of the data (image, sound, and other markers) captured during the screening process as well as the overall quality of the results and the care pathway downstream leading to clinical efficiency and diagnostic accuracy throughout. For example, radiology practices that involve AI are also likely to reduce the number of false-positive exams, eliminating unnecessary costs and patient anxiety associated with follow-up procedures.

AI-aided screening and diagnostics tools can not only improve disease diagnosis but also provide a comprehensive care approach that can help healthcare professionals in clinical decision-making by analyzing symptoms, suggesting personalized treatments, and predicting risk. AI is a powerful tool that can be skillfully deployed to make health systems affordable and accessible.

Smartphones based AI technologies- promising solution

Smartphones based AI technologies present great potential in enabling low-cost screening of various medical conditions by collecting data from smartphone sensors and using machine learning algorithms to analyze data. Frontline health workers can easily use a smartphone to screen for a variety of medical conditions by using AI.

For example, for people with hypertension, cardiac disease, and arrhythmia, it's crucial to regularly check vital signs including blood pressure and heart rate. Many methods have been developed that can precisely estimate these vital indicators using smartphone image sensors. Smartphone applications may analyze recordings of a patient's fingertip using photoplethysmography technology to calculate blood pressure, heart rate, and heart rate variability, resulting in readings that are equivalent to those acquired with conventional cuff devices and electrocardiograms.

These platforms, once thoroughly validated, could increase screening accessibility, and improve early treatment to avoid heart disease and strokes. Smartphone AI technologies could enable non-invasive screening and have the potential to complete initial assessments that can be used to refer patients to specialized care.

Some promising work is already happening

In India, Smart Scope by Periwinkle Technologies carries out digital cervical imaging with AI-enabled assessment of the risk level and generates color-coded reports at the point of care. Qure.ai uses AI for reading chest x-rays. Their qXR tool aids in the detection of multiple abnormal findings on a chest X-ray in?less than a minute. qXR can detect abnormalities in the lungs, pleura, mediastinum, bones, diaphragm, and heart. Niramai Health Analytix developed Thermalytix which is a computer-aided diagnostic engine that is powered by AI. The solution uses a high-resolution thermal sensing device and a cloud-hosted analytics solution for analyzing thermal images to screen for breast cancer. Swaasa uses machine learning models to analyze cough sounds in combination with other information like temperature, oxygen saturation, and symptoms to assess the performance of the lungs. EzeCheck by EzeRx is a non-invasive AI-enabled portable device that can screen for blood glucose, and anemia, and can predict kidney, liver, and lung problems.

Challenges and way forward

It is quite evident now that AI in population health screening can improve access to diagnostics, improve the quality of clinically relevant data, and will immensely support clinical decisions. But its implementation and operations are not an easy feat. High-quality labeled datasets are required to train supervised machine learning algorithms. Algorithms trained on datasets collected in high-income countries might not be generalized for deployment in other settings. Building models using non-representative data can introduce algorithmic bias, particularly towards minority ethnic groups.

A variety of ethical implications around the use of AI in healthcare is another hurdle. Data ethics is the foundation of AI, and its key areas include informed consent, privacy and data protection, ownership, objectivity, and transparency. Healthcare decisions have been made almost exclusively by humans in the past, and the use of smart machines raises issues of accountability, transparency, permission, and privacy.

To overcome the challenges posed by AI, it is vital to collect data from the target population to train and assess model performance. A key strategy for facilitating quick development and replication could be the creation of shared data repositories where researchers can externally assess and improve models.

To prevent the negative effects of AI use in healthcare, including population screening, it is crucial that healthcare organizations, as well as political and regulatory agencies, build structures to monitor major issues, respond responsibly, and establish governance procedures. It is necessary for AI systems to receive regulatory approval, be integrated with EHR systems, be sufficiently standardized so that similar products function, similarly, be taught to doctors, be paid for by public or private organizations, and be improved over time in the field.