[Ed Note: The following post is part of the TLF Editorial Board Test 2020-21. It has been authored by Yashashwini Santuka, a second year student of NALSAR University of Law.]
Advanced systems of healthcare are imperative to the growth of countries, their economies and the well-being of its people. However, developing countries like India are still in the process of adapting to emerging technology in public healthcare due to its resource-constrained setting. The use of Artificial Intelligence (AI) in this scenario is rapidly spreading in public health. Effective deployment and adapting to its unique features to transform public health completely might take longer due to the systemic disparities observed in the country. While AI holds promise for the health systems, its uniform implementation may pose an issue to traditional patient care systems, patients’ safety, safety of their private medical records, and affordability. Such a situation requires regulators to take a systemic view of the healthcare industry, and possibly pre-empt the potential impact of the use and regulation of AI. This article explores the contextual limitations of the healthcare industry in India concerning the regulation of technology and AI.
Artificial Intelligence in India
The public healthcare systems in India remain overburdened as they serve as the only point of accessible healthcare for the country’s rural population, perhaps due to heavily subsidised rates or free services. While the Ayushman Bharat (AB-PMJAY) policy and its aim of “health for all” has helped the system, it still requires the aid of advanced technology to support healthcare delivery and its efficiency.
Artificial Intelligence is increasingly demonstrated by machines through the performance of various cognitive tasks by acquiring and applying knowledge, as a result of the convergence of multiple technologies and algorithms. In medicine, AI operates in an ecosystem of health data used to train machines to diagnose, detect, predict diseases, suggest medication, and perform complex medical tasks. It is presented as one step further and more efficient in terms of detection of diseases through factors that may not be discernible to the clinician’s eye easily. Such use has been documented in Electronic Health Record (EHR) systems abroad, which have used machine learning to detect significant medical details through the text, analyze, even use predictive analysis to warn doctors about high-risk conditions, etc. Additionally, it also helps analyze data for a larger demographic concerning pandemic preparedness and response measures.
India, through its National Health Policy (2017), has made clear its agenda to digitise the healthcare industry, which was further strengthened by the vision presented in the National Digital Health Blueprint (2019) to identify building blocks in the foundational technology to support the expansive application of this technology in a simple and varied manner. Healthcare is embracing the digital health innovation, in terms of the growing use of the technology to respond to critical health challenges.
Challenges in the Use and Regulation of AI
Despite the ever-growing need for the use of AI in the Indian healthcare regime and opportunities to carry out the same there are multiple roadblocks in the effective implementation, use and regulation of technology. The use of AI has permeated through and is available in speciality and tertiary care hospitals. However, few AI-related solutions reach primary care centres perhaps due to the high cost of development. The widespread use is also prevented due to its complexity, which creates the perception of a ‘black box’ as certain principles are deemed inaccessible – and hence untrustworthy – by users. Due to its use of several data points and machine algorithms, its working is difficult to comprehend without human intervention and aid. This heavy reliance on pre-existing data sets and recurring patterns makes AI vulnerable to replicate discrimination against disadvantaged groups that could otherwise be ruled out by clinicians.
The country’s National Strategy for AI sets precedent for AI development through the institute of Centre for Research Excellence (COREs) for research, while acknowledging challenges around issues of data safety, privacy, data integrity and technical resource capacity. In a bid for higher efficacy of public health systems, a certain degree of caution is substantial to stay proactive in terms of pre-empting the difficulties that are bound to arise, such as data safety.
While the system is presented with several challenges relating to regulation of AI, a few significant issues arise at the pre-regulatory stage. First arises the conundrum of data – its ownership and data-sharing between private and public healthcare systems seems to hamper the use of these technologies. The government, in this case owns large swathes of data – from public health facilities and national programmes – but it lacks completeness and accuracy. In this case, aggregation of minor inaccuracies, such as spelling mistakes can culminate into major misinterpretations hampering representativeness of data sets used for training and creating biases. Therefore, enabling digitisation of clinical transactions that the citizens partake, is integral to prevent severe social fall-outs due to the biases. Beyond this, the private institutions with digital systems face issues concerning patients’ data due to lack of interoperability between the records of the government and private institutions. The health systems due to this remain fragmented and incomplete.
Secondly, under the Information Technology (Reasonable security practices and procedures and sensitive personal data or information) Rules 2011, health data is listed as sensitive personal information which necessitates stronger privacy and security measures. The use of anonymisation and de-identification techniques are used to preserve privacy of the data, by removing all identification points from the data and prepare it for sharing for training AI systems. However, this is not absolute. Annotation of health data is yet another process required by machines to learn and draw patterns. These processes are, however, investment-heavy and cumbersome that make the process expensive and inclusive such that it is difficult for an enterprise to enter the regime.
Thirdly, the nature of algorithms is known to be dynamic and with the tedious nature of standard regulatory measures, it is not hard to imagine situations of release of an improved product after an earlier approved version has been marketed. The US for such situations has adopted a total product life cycle (TPLC) approach, which allows the rapid cycle of product improvement while mandating the pre-market submission for changes that affect safety and effectiveness. This remains an impediment in India in the absence of a solid policy to facilitate such programs.
Fourthly, the regulations in place, their agility, and responsiveness, have a deep and direct impact on the adoption of innovations. These need to be regularly fine-tuned to enable optimal innovations, when handling healthcare expenditure. India’s medical device market has grown in leaps and will continue to do so, but its regulatory framework remains a hurdle due to its deficiencies. It is thus, imperative to ascertain the use of technological mechanisms and legally permit it, to check if its probable benefits outweigh the probable risks involved, and ensuring these products are easily made available. Medical device regulation proved to be a major hurdle. It was only in 2017 when medical devices were categorised separate from drugs/pharmaceuticals, software was still not included as mentioned in its earlier draft. It was only in February 2020, through two notifications, software or an accessory to be used as a medical device under the purview of regulation.
Conclusion
India, however, still lags behind its international counterparts in terms of regulations to promote innovations. As the country moves through an advanced stage of regulatory measures introducing a wide ambit of medical products to be used in the health system, its regulators need to assess the needs of multiple stakeholders, including that of protecting patients’ interest economically through check on prices i.e. affordability and their health through various stages of pre and post-market regulations. India can adapt the pre-existing reference regulations (International Medical Device Regulators Forum or IMDRF) and mould its policy-making process to make it more participatory, for the creation of a synergistic ecosystem for AI in healthcare.