The integration of artificial intelligence (AI) into healthcare is poised to bring significant changes to the landscape of medical treatment and patient care. However, experts underscore the importance of inclusivity and representation in the data used to build these AI models to ensure their effectiveness across diverse populations. Dr. Georgi Chaltikyan, the program director for the Master of Digital Health at the Deggendorf Institute of Technology, highlighted these critical aspects in an interview with Health Tech World.
Dr. Chaltikyan emphasises that the quality of AI depends heavily on the datasets used to train the models. When these datasets are skewed or overrepresent certain populations, it can lead to biases that compromise the AI's utility in real-world applications. Notably, underrepresented groups, particularly in low- and middle-income countries, often suffer from inadequate digital health data, primarily due to a lack of robust digital infrastructure. "Good AI requires models trained on data that accurately reflect the diverse patient population they are designed to serve," Dr. Chaltikyan stated.
The importance of inclusivity in data cannot be overstated, as health outcomes can differ significantly due to variations in genetics, treatment responses, and disease progression among different demographic groups. "It's also about equity," Dr. Chaltikyan added, cautioning against creating a digital divide, which could exacerbate existing disparities in healthcare access and treatment. This divide may stem from various factors, including generational gaps in digital literacy, where younger individuals are more adept at utilising new technologies compared to their older counterparts.
Dr. Chaltikyan elaborates that the overarching aim of healthcare digitalisation is to place the patient at the centre of their own care. He asserts that it is imperative that not only healthcare specialists but also patients engage with digital health innovations. The active participation of patients in their health journey will significantly contribute to the adoption of digital health tools such as telehealth applications and sensor monitoring technologies.
For successful digitalisation, Dr. Chaltikyan identifies three key stakeholders: developers, healthcare professionals, and patients. He notes that collaboration among these groups will be essential for the wide-scale implementation of digital health solutions.
A significant aspect of this digitalisation vision includes the "10P Health" concept, which expands upon the earlier "4P" model of personalised medicine pioneered by Dr. Leroy Hood. The ten principles encompass Predictive, Preventive, Personalised, Precision, Participatory, Pertinent, Proactive, Pervasive, Permanent, and Platform-based health and wellness. Dr. Chaltikyan explained that the essence of "10P Health" revolves around precision and personalised health, which tailors treatments to the individual rather than relying solely on standardised protocols.
He draws a compelling analogy between AI in healthcare and an autonomous vehicle's onboard computer. "Imagine your body is the car, and the digital wellness platform is the onboard computer," Dr. Chaltikyan said. This platform assists users by providing insights and recommendations based on their health data, flagging potential risks, and allowing for preventive care that can enhance health outcomes in the long term.
However, he notes that the effectiveness of these AI systems relies on high-quality models backed by inclusive data sets. "All of this, of course, requires high quality models, and high quality models require data, and data, ideally, must be inclusive," he concluded. The focus on integrating diverse data into AI systems is not merely a technical requirement but a fundamental necessity in reshaping healthcare for the betterment of all populations.
Source: Noah Wire Services