The COVID-19 outbreak caught public health agencies, and virtually everyone else, unaware. Agencies lacked the strategic intelligence that could have helped them prepare and manage the unprecedented disruption caused by the virus. The current infection wave has generated a substantial amount of data. While it may not be enough to predict the timing of the reoccurrence, this information can help agencies be better prepared for any future waves.
Close collaboration among CIOs and Chief Analytics Officers (CAOs), Chief Data Officers (CDOs), agency heads (state and federal organizations, health and human services), healthcare providers, and managed care organizations, supported by an advanced AI and data science platform, can help government at all levels adopt preventive protocols to achieve optimal outcomes.
COVID-19 data and analytics challenges
Constantly evolving and noisy data, outliers, and the lack of historical information have made it difficult to build and train effective COVID-19 AI models. Initial forecasts have been unreliable, with most predicting an even worse case scenario than what we are currently experiencing.
The inability of public health agencies to effectively collaborate with their partner agencies and departments hasn’t helped either. Data and intelligence on COVID-19 if consolidated might have helped manage the epidemic, but remained stuck in silos.
Lack of effective data governance, data stewardship, and privacy and security concerns combined with an incoherent analytics strategy and interoperability issues made it difficult to collect, collate, standardize, analyze, and disseminate COVID-19 intelligence across the health care ecosystem.
Solving the data and analytics challenges for the next COVID wave
Fast and accurate tracing of the pathogen reproducibility is key to saving lives in any future waves of COVID. The ability to predict the virus epicenter and the spatial direction of spread over time, correlated with information such as transportation and population density, can help officials decide the lengths/types of shutdown (immediate and prolonged, multiple and short) and the social distancing programs that need to be implemented to contain the epidemic.
Identifying the most vulnerable population and predicting the case fatality rate will be crucial in demand forecasting and optimization of medical resource allocation. Identifying the traits of the exposed people who were asymptomatic or had mild cases compared to the rest who required hospitalization or expired will help with adequate capacity planning. It will be equally important to understand what societal and economic/behavioral composites will aggravate the COVID situation.
Limitations of the past are not necessarily limitations of the future. Healthcare agencies need to augment their COVID-19 data set with available historical patient records including medical and pharmacy claims, demographics, disease registry, vital records, EMR data, lab results, and social determinants of health attributes to create a comprehensive population health data registry and longitudinal records at the individual level. Different organizations have these data sets. They all need to work together to build a comprehensive open data hub.
Use of next-generation data science platforms can help analyze all this data in real time for sophisticated descriptive, diagnostic, spatial and predictive analytics, enabling agencies to take a proactive, risk-based approach for population and individual health management. Following are a few guiding principles to adopt a next-generation data science platform to analyze data meaningfully and at scale:
- Cloud native to ensure platform scalability and elasticity
- Supports polymorphic data and polyglot storage
- Open architecture for agility and collaboration
- Unified data experience and ease of cross-agency data exchange
- Facilitates data exploration and encourages effective data stewardship
- Facilitates driverless data lifecycle management
- Invokes self-service analytics to empower data democratization
- Supports automated machine learning for rapid AI model development
- Disseminates AI-driven, next-best-action-oriented intelligence to all stakeholders in real time
CIOs must reimagine their role and become a champion of data and intelligence within the organization. Their vision and direction will be critical to ensure investments in scaling up of the existing data infrastructure to meet the challenges of the next wave. This means advocating for a data-driven culture across the organization, establishing security practices, assessing risk, designing data-management policies, institutionalizing best practices in data science model enablement, and encouraging collaboration between business leaders and interagency stakeholders.
Infosys Public Services and AWS have partnered to develop a self-service, AI-based advanced data analytics solution (Infosys DIA- Data Insights Actions Platform) that can help CIOs do the above. The platform is built using best-of-breed data science tools and AI techniques, and is designed as “Analytics in a Box.” The platform enables agencies to easily aggregate, manage, and analyze multi-modal data to generate the right intelligence and deliver the right interventions to the right people, at the right time, for improved outcomes.
To disseminate rapidly changing COVID information, the DIA platform has been extended with AI-based digital health advisors. These digital advisors can access up-to-date information from reliable sources like the CDC and the WHO, and leverage various artificial intelligence techniques such as natural language understanding and processing, machine learning, neural ontology, and emotional intelligence to engage people meaningfully, address their queries accurately, and enable staff to focus on strategic initiatives.
Future waves of the novel coronavirus are a certainty. Effective vaccines are still likely to be 12 to 18 months away from being available. No evidence exists yet of any herd immunity towards the virus. Health care system CIOs must act now to strengthen their data and analytics systems because the right data sets (an ideal combination of clinical and non-non-clinical), the right insights, and the right actions (pharmaceutical and non-pharmaceutical) can reduce the spread of the virus and mitigate its impact on public health, business operations, and the economy.