About the author:
Jivya Lamba is deeply passionate about science and technology. She finds fulfilment in assisting others and has contributed to supporting underprivileged girls in India by authoring a PCOS information handout and providing them with bicycles to commute to school. Currently, Jivya is in the 12th grade at the Modern School Vasant Vihar in New Delhi. She aspires to pursue a career intersecting Chemistry and Data Science.
Abstract
Diabetes, a prevalent and chronic health condition characterized by elevated blood glucose levels poses a significant global health challenge. Early detection and prevention play pivotal roles in mitigating the impact of diabetes on individuals and healthcare systems. In recent years, the application of machine learning has garnered substantial attention for its potential to revolutionize diabetes management. This research article explores the utilization of machine learning techniques for early detection and prevention within the Indian healthcare sector.
Introduction
Diabetes is a medical condition characterized by two primary mechanisms: insufficient insulin production by the pancreas or ineffective utilization of the produced insulin by the body. Insulin, a vital hormone, is responsible for the regulation of blood glucose levels. Hyperglycemia, often referred to as elevated blood glucose or heightened blood sugar, is a common consequence of unmanaged diabetes. Over time, hyperglycemia can result in severe damage to numerous physiological systems.
According to a World Health Organization’s report, in 2014, approximately 8.5% of adults aged 18 and older were afflicted by diabetes. In the year 2019, diabetes directly contributed to 1.5 million fatalities, constituting 48% of all deaths attributable to this disease (WHO, 2019). According to the growing morbidity in recent years, in 2040, the world’s diabetic patients will reach 642 million, which means that one of the ten adults in the future is suffering from diabetes.
Between the years 2000 and 2019, there was an observed 3% rise in age-standardized mortality rates linked to diabetes. This increase was even more pronounced in lower-middle-income nations, where diabetes-related mortality surged by 13%. In contrast, there has been a noteworthy global decrease of 22% in the likelihood of succumbing to any one of the four primary noncommunicable diseases (comprising cardiovascular diseases, cancer, chronic respiratory diseases, or diabetes) between the ages of 30 and 70 years. This decline signifies recent progress in addressing these noncommunicable diseases on a global scale (Verma, Khanna and Mehta, 2012).
Machine Learning and the Healthcare Sector
The past century has witnessed a substantial increase in average life expectancy, driven by technological advancements. Emerging technologies such as Artificial Intelligence (AI) and ML herald a new era for healthcare. Through computational prowess, even the minutest aspects of medical operations can be optimized to near perfection. Although ML has already made inroads into healthcare, its untapped potential for future applications remains vast. For this paper, the terms artificial intelligence (AI) and machine learning (ML) may be used interchangeably.
Machine Learning (ML) encompasses a diverse set of statistical techniques enabling computers to acquire knowledge through experience, devoid of explicit programming. With the rapid development of machine learning, machine learning has been applied to many aspects of medical health (Zhang et al, 2018). This acquisition of knowledge typically entails modifying algorithmic behavior (Bush, 2018). ML can, for instance, recognize faces by studying a dataset of photographs depicting various individuals.
Among the various sectors poised to harness the potential of ML, healthcare stands out. The healthcare industry has consistently embraced cutting-edge technologies, akin to their adoption in business and e-commerce sectors. The potential applications of ML in healthcare appear limitless. With its pioneering applications, ML is reshaping the healthcare landscape, enhancing automation and intelligent decision-making across primary, tertiary patient care, and public healthcare systems (Varma, Manoj and Panda, 2019). This transformation has the potential to improve the quality of life for billions of individuals worldwide.
For instance, ML could be a promising tool to maximize new-onset diabetes prediction than conventional statistics models, reporting an accuracy variable from 71% to 94% and exploiting a dataset composed of a minimum of 3700 patients up to a maximum of 2 million (Agliata et al, 2023).
ML technologies also hold extensive promise for optimizing clinical trial research. By employing advanced predictive analytics to evaluate potential clinical trial participants, medical professionals can process a broader range of data, reducing both costs and time associated with medical assessments.
ML also offers solutions to enhance clinical trial efficiency, including determining optimal sample sizes for enhanced efficacy and mitigating data errors through Electronic Health Records (EHRs). ML’s potential extends to research and clinical trials, where predictive studies based on ML can identify potential clinical trial participants by drawing insights from diverse data sources such as previous medical visits and social media. Real-time data access and trial management further streamline the investigation process, optimizing sample sizes and reducing data-related errors. Addressing the shortage of well-trained radiologists, ML can aid in diagnosing and analyzing medical imaging data efficiently. Today, electronic medical imaging data abounds, offering a rich dataset for analysis. ML algorithms can examine imaging data akin to skilled radiologists, detecting anomalies, lesions, tumors, and even brain bleeding. Furthermore, ML facilitates personalized, dynamic therapies by merging individual health data with predictive analytics.
Machine Learning and Diabetes
Diabetes has emerged as a significant health concern in South-East Asia, where an estimated 23 million individuals currently grapple with the condition, representing one-sixth of the global diabetic population. Notably, India boasts the largest diabetic populace and one of the world’s highest diabetes prevalence rates. Projections indicate an increase from 19.4 million in 1995 to a staggering 80.9 million by 2030. Particularly concerning is the projection of the most significant increases in diabetes cases occurring within economically productive age groups in developing nations (Kaur and Kumari, 2020). Given the current elevated mortality and morbidity rates associated with diabetes, this poses a tangible threat to the economic productivity of countries like India.
Presently, major diabetes health initiatives primarily focus on integrating diabetes healthcare into existing disease-prevention programs, such as those targeting heart disease and hypertension, which share similar risk factors. These initiatives aim to establish active educational programs for diagnosed patients regarding their risk factors and well-structured referral systems to specialists when necessary. Several prominent studies have unmistakably demonstrated a positive correlation between effective disease management and a reduction in disease burden. These findings underscore the potential for mitigating at least a portion of the economic and social burdens imposed by diabetes through appropriate treatment regimens. However, there persists a noticeable lack of awareness among policymakers and healthcare strategists regarding its gravity.
Machine learning, nestled within the realm of artificial intelligence, empowers us with potent tools for the early detection of diabetes (Kalyankar, Poojara, & Dharwadker, 2017) through Predictive Models. Machine learning algorithms harness the prowess to scrutinize extensive datasets encompassing medical records and genetic insights, enabling the prediction of an individual’s susceptibility to diabetes. These models encompass a wide spectrum of risk factors, thus facilitating proactive interventions and lifestyle modifications.
Machine learning systems aid healthcare practitioners in the more precise and expeditious diagnosis of diabetes and its subtypes. This is achieved through a meticulous analysis of symptoms, laboratory findings, and comprehensive medical histories. By dissecting individual patient data, machine learning facilitates the customization of prevention strategies to tackle specific risk factors (Nithya and Ilango, 2017). These personalized interventions span a broad spectrum, encompassing tailored dietary recommendations and personalized exercise plans. Machine learning applications motivate individuals to embrace and sustain healthier lifestyles by delivering behavioral feedback, timely reminders, and enticing incentives.
Policy Recommendations
By identifying individuals on the cusp of prediabetes, a precursor to Type 2 diabetes, machine learning empowers early interventions to impede the progression to full-fledged diabetes. In synergy with an evolving understanding of diabetes types, machine learning emerges as a formidable ally in the quest to revolutionize diabetes detection and prevention strategies, promising to enhance the quality of life for millions worldwide (Saru and Subashree, 2019). The fusion of clinical expertise and machine learning’s analytical prowess holds the potential to usher in a new era of precision medicine in diabetes care, ensuring earlier interventions, reduced complications, and improved outcomes for patients across the globe.
In its national AI strategy, India has adopted a distinctive approach, emphasizing the utilization of AI not solely for economic advancement but also for fostering social inclusion (Parry & Aneja, 2020). The strategy, conceived and articulated by NITI Aayog, the government’s think tank, is termed #AIforAll. Consequently, the overarching objectives of the strategy include enhancing skills for quality employment: One of its central goals is to equip the Indian populace with the requisite skills to secure high-quality employment opportunities. Additionally, the strategy underscores investment in research and sectors that have the potential to maximize economic growth while simultaneously generating substantial social impact (Sarwar et al, 2018). Furthermore, the goal of scaling up Indian-developed AI solutions to benefit other developing nations worldwide is also laid out.
NITI Aayog published India’s AI strategy document on June 4, 2018, and in its formulation, NITI Aayog engaged in collaborative processes involving experts and stakeholders. These processes encompassed the development of AI projects in various domains supported by comprehensive evidence and the design of a strategy aimed at nurturing a vibrant AI ecosystem within India.
Recognizing AI as a transformative technology, NITI Aayog introduced the branding concept of #AIforAll to facilitate the widespread adoption of AI in India. This initiative aligns with India’s aspirations to assume a leadership role in AI development and underscores the strategy’s core objective of harnessing AI for inclusive socio-economic growth. Ultimately, the strategy positions India at the forefront of AI technology development, intending to serve as a hub for emerging and developing economies (Pallathadka et al, 2023). It examines the existing AI development ecosystem in India, identifies potential sectors for AI integration, assesses research and development capabilities, and outlines the roadmap for the future.
NITI Aayog’s policy recommendations within the strategy encompass over 30 facets. These recommendations advocate investment in scientific research, the promotion of reskilling and training initiatives, acceleration of AI adoption across the value chain, and the establishment of ethical, privacy, and security standards in AI applications (Islek et al, 2020). A notable flagship initiative involves the creation of two-tiered integrated AI research centres. The first tier comprises Centres of Research Excellence in AI (COREs) dedicated to fundamental research, while the second tier encompasses International Centres for Transformational AI (ICTAIs), which focus on the development of AI-based applications in domains of societal significance.
In alignment with the strategy, NITI Aayog identifies priority sectors as healthcare, agriculture, education, smart cities, and smart mobility. The strategy further recommends the establishment of Ethics Councils at each CORE and ICTAI, the formulation of sector-specific guidelines on privacy, security, and ethics, the creation of a National AI Marketplace to streamline market discovery and data collection processes, and the initiation of workforce development initiatives to bolster AI skills across the nation (Thotad, Bharamagoudar and Anami, 2023).
India, often referred to as a “country with no records,” although not entirely accurate highlights a prevailing issue within the medical sector – the deficiency in comprehensive record-keeping practices. This deficiency presents an enormous opportunity: by harnessing the entirety of our medical data in a structured and accessible format, AI and ML can be leveraged to derive tailored insights and solutions specifically tailored to our diverse population. AI has the potential to facilitate personalized treatment approaches across multiple conditions, enable efficient healthcare delivery, and lead to a more connected healthcare ecosystem. Data is omnipresent in healthcare and, as we harvest more of it, our AI and machine learning capabilities will be able to grow exponentially (Siddiqui, 2021).
Conclusion
In conclusion, the effectiveness of Artificial Intelligence and Machine Learning hinges upon the quality and availability of the data utilized for their training and application. Realizing this potential requires a concerted and collaborative effort involving not only the government but also organizations such as the Endocrine Society of India. These entities should take the lead in initiating data collection and research initiatives that would serve as the bedrock for future advancements in healthcare powered by AI and ML. By addressing the dearth of medical records and facilitating the integration of these technologies into our healthcare system, more effective, personalized, and equitable healthcare solutions can be generated for India.
References
- Bharadwaj, Hemantha Krishna, Aayush Agarwal, Vinay Chamola, Naga Rajiv Lakkaniga, Vikas Hassija, Mohsen Guizani, and Biplab Sikdar. “A review on the role of machine learning in enabling IoT based healthcare applications.” IEEE Access 9 (2021): 38859-38890.
- Ghosh, Pronab, Sami Azam, Asif Karim, Mehedi Hassan, Kuber Roy, and Mirjam Jonkman. “A comparative study of different machine learning tools in detecting diabetes.” Procedia Computer Science 192 (2021): 467-477.
- Islek, Duygu, Mary Beth Weber, Anjana Ranjit Mohan, Viswanathan Mohan, Lisa R. Staimez, Ranjani Harish, KM Venkat Narayan, Michael Laxy, and Mohammed K. Ali. “Cost-effectiveness of a stepwise approach vs standard care for diabetes prevention in India.” JAMA Network Open 3, no. 7 (2020): e207539-e207539.
- Jayatilake, Senerath Mudalige Don Alexis Chinthaka, and Gamage Upeksha Ganegoda. “Involvement of machine learning tools in healthcare decision making.” Journal of Healthcare Engineering 2021 (2021).
- Kaur, Harleen, and Vinita Kumari. “Predictive modelling and analytics for diabetes using a machine learning approach.” Applied Computing and Informatics 18, no. 1/2 (2020): 90-100.
- Nithya, B., and V. Ilango. “Predictive analytics in health care using machine learning tools and techniques.” In 2017 International Conference on Intelligent Computing and Control Systems (ICICCS), pp. 492-499. IEEE, 2017.
- Pallathadka, Harikumar, Malik Mustafa, Domenic T. Sanchez, Guna Sekhar Sajja, Sanjeev Gour, and Mohd Naved. “Impact of machine learning on management, healthcare and agriculture.” Materials Today: Proceedings 80 (2023): 2803-2806.
- Saru, S., and S. Subashree. “Analysis and prediction of diabetes using machine learning.” International Journal of Emerging Technology and Innovative Engineering 5, no. 4 (2019).
- Sarwar, Muhammad Azeem, Nasir Kamal, Wajeeha Hamid, and Munam Ali Shah. “Prediction of diabetes using machine learning algorithms in healthcare.” In 24th international conference on automation and computing (ICAC), pp. 1-6. IEEE, 2018.
- Thotad, Puneeth N., Geeta R. Bharamagoudar, and Basavaraj S. Anami. “Diabetes disease detection and classification on Indian demographic and health survey data using machine learning methods.” Diabetes & Metabolic Syndrome: Clinical Research & Reviews 17, no. 1 (2023): 102690.
- Varma, Kucharlapati Manoj, and B.S. Panda. “Comparative analysis of Predicting Diabetes Using Machine Learning Techniques.” J. Emerg. Technol. Innov. Res 6 (2019): 522-530.
- Verma, Ramesh, Pardeep Khanna, and Bharti Mehta. “National programme on prevention and control of diabetes in India: Need to focus.” The Australasian Medical Journal 5, no. 6 (2012): 310.