Management Perspectives

Managing personal information is a challenge in an environment marked by high digitalisation and involvement of technology in all aspects of an individual’s life. It is all the more so in the case of healthcare data, one of the most critical aspects of an individual’s personal information and an important component of the data exchange that occurs in the health ecosystem.

The pervasive healthcare system demands that information is available at the right time for the right use. However, an individual’s health information is spread across his or her entire life cycle, and is usually stored in outpatient clinics and hospitals, at the workplace and home; or is available with health insurers. With different stakeholders using different storage devices and following different documentation patterns and recording systems, accessibility of health information often becomes difficult.

This points to the importance of integrating this multi-source information to create intelligent decision-making systems. Also, utilisation of such data demands a complete understanding of the degree to which technology can contribute to managing it across the spectrum.

Application of emerging technologies, like predictive analytics and machine learning, by hospitals and health insurers holds immense promise in resolving this issue; it can also help address problems related to reducing the cost of managing health data and insurance claim adjudication errors. Adopting these technologies within a defined regulatory framework can benefit patients too—through better understanding of disease patterns, improved treatment protocols, and secure data privacy norms. This is the principal goal of machine learning systems. Based on health records maintained by individuals, insurers can drive predictive analytics for making preventive healthcare interventions.

New technologies, better access
Insurance is an important financing mechanism to improve access to quality healthcare. In emerging economies like India, the increasing burden of lifestyle diseases has made its role vital in providing preventive and wellness solutions as opposed to curative care

In developed health insurance markets like the United States too, issues of poor quality and high costs necessitate technology-driven healthcare outcomes. Contrary to general belief, poor quality and also non-uniformity of coding systems plague the system in the US too. However, the Health Information Technology for Economic and Clinical Health Act of 2009 significantly increased the adoption of electronic health records (EHRs) there. Similarly, in India, the Clinical Establishment Act, 2010 made it mandatory for healthcare providers to work towards EHRs to increase the efficiency of healthcare delivery. Although the scenarios are different, utilisation of predictive modelling for reducing the burden of diseases is imperative in both countries. Predictive analysis incorporates a variety of techniques like data mining, statistics, and game theory. It uses current as well as past data with statistical or other analytical models and methods to determine or predict future events; it also enables healthcare providers to develop focused solutions for patient needs.

Figure 1 depicts the sources of clinical information generation and storage for a patient: Information available at an outpatient clinic, hospital, workplace, and home; or with the insurer, may have varying levels of complexity. The enormous amount of data generated can be synthesised using regression-based data mining techniques to predict future healthcare event patterns for an individual. This would result in not only lower treatment costs but also better allocation of healthcare resources.

Plugging loopholes
Health insurance claims data is a rich source of information for developing predictive models for policy-making and treatment interventions. However, its complexity poses numerous challenges to researchers and policymakers. Errors leading to wrong payments are a major cause for increasing healthcare costs across the world. Machine learning techniques could help in reducing such errors, which would help in not only improving the quality of information captured but also optimising operational costs. The claims data can then be studied to understand the patterns of clinical data and related disease outcomes. Standard algorithms can then be developed to review future claims and determine the outliers vis-à-vis expected disease outcomes.

Figure 2 represents a health claims processing model that uses machine learning, which can help clinicians make better and more accurate decisions. The benefit configurator is a modular technology platform that brings together financial due diligence of a health insurance product, claims processes, and rules like waiting periods for specific diseases, differential access to hospital networks, and so on. These product rules are applied based on demographic parameters like age, gender, or income limits of the insured member. Personal health records comprise clinical history details gathered by the insurance firm. These three aspects are integrated during claims processing to understand the risk propensity of the insured member.

Clinical information is created under different healthcare settings, and patients are required to assimilate all data in a manner that can be used for making meaningful interpretations. Clinicians too need to put in extra effort to interpret and utilise information obtained after integrating information from diverse sources.

The large amounts of data generated at each patient interface with the healthcare system needs special attention. Machine learning techniques can help to a great extent in identifying trends and patterns in it for further analysis.

Utilising the domain knowledge of experts as well as information generated through machine learning techniques, predictive healthcare models can be created for different diseases. Similar predictive models are already being used by insurers to reduce leakages and wastage in their internal processes through correct coding of diseases.

Driving transformation
Predictive analytics and machine learning can also address other problems in the healthcare community. For instance, it could help in long-term care for ailments like cancer and diabetes, which need more attention from a health information management perspective since clinical data needs to be preserved for a longer period. However, it is highly important to take care of legal issues related to privacy and explore the extent to which predictive models or machine learning can compromise the security of such data. With proper security measures in place, predictive models supported by machine learning can transform healthcare markets in developing economies like India as well as developed markets like the US