Strategies for Using AI to Drive Population Health and Address SDOH

The following is a guest article by Pravin Pant, Vice President, Advanced Analytics at ZeOmega

As much as 80% of an individual’s health is determined by social and economic conditions. Social determinants of health (SDOH) are pivotal in shaping population health, providing care teams with a comprehensive understanding of the factors affecting their patients.

Health plans and providers recognize that challenges such as food insecurity or inadequate housing hinder optimal health outcomes without additional support extending beyond clinical care. Therefore, healthcare organizations must implement effective strategies to gather SDOH data and apply targeted interventions to improve population health outcomes. By adopting the right approaches, organizations can address SDOH and ensure patients receive the care they need, while simultaneously reducing costs, emergency services utilization, and employee burnout.

To build effective SDOH strategies, payers and providers must leverage technology such as artificial intelligence (predictive and GenAI) to integrate diverse data from both public and private sources, including assessments, surveys, and social media. It is important to collect community-level SDOH information from publicly available databases, mapped to the census tract level, and combine it with experiential claims data and personal assessments to create comprehensive digital member profiles. These profiles are critical for machine learning (ML) and predictive next steps.

When health information is collected from a variety of databases, it becomes invaluable to care teams, providing visibility into SDOH risks alongside behavioral and clinical risks within a population cohort or an individual. Insights may identify the probability of homelessness, possible chronic conditions, and other risks such as opioid overdose. The information, when made available at the point of care, allows clinicians to make proper care assessments and transition of care. Addressing risks early enhances the quality of care, while unaddressed risks can lead to a decline in overall care quality.

By leveraging GenAI, care teams can adopt a holistic, whole-person care approach that addresses medical, social, and behavioral health. GenAI also guides the next best preventive care strategies and their implementation for improved outcomes. But generally, the main idea is that GenAI can help alert care teams to multiple health issues that can be addressed and prevented simultaneously. From the health plan perspective, GenAI helps care teams tailor interventions for current and predicted issues. For example, research studies and data show that people at high risk for education and transportation have much higher poverty levels.

In addition, by identifying high SDOH risk members early, healthcare organizations can be proactive in addressing the non-clinical/social needs so that when/if a member develops clinical risk such as diabetes, both the member and plan will be better able to manage the challenge. However, it is important to remember that predictions are only as good as the data they are based on. Therefore, having a sound interoperability strategy in place coupled with an SDOH strategy is critical. Interoperability plays a major role in successfully managing and exchanging data between systems, which in turn supports the generation of effective predictive models that health plans need to develop and deploy strong intervention approaches.

SDOH strategies and solutions can be enhanced with a predictive next step, meaning clinical users can include the comprehensive member profile generated from EMR information and member behaviors. By applying ML methodologies, individual members’ or cohorts’ needs can be met, helping clinical organizations be proactive about addressing the nonclinical/social needs through automated next best action and intelligent care workflows. Because these insights are more precise, they might suggest new kinds of health initiatives that may have been historically overlooked or never considered (e.g., sending a mobile clinic to certain neighborhoods where transportation options are limited or starting a “how to cook” program in areas that face food insecurity).

GenAI, such as GPT-4 framework, can be critical when it comes to understanding individuals’ circumstances their doctors and health plans may not be aware of. It can help provide visibility into a person’s SDOH needs that were unable to be identified during direct interactions and recommend the precise downstream action for a member or cohort. Furthermore, GenAI frameworks help validate and justify why a person will need certain SDOH help and auto-approval. For example, if a physician needs to send a prior authorization for an MRI, the system will also flag that the patient will need transportation to complete the scheduled MRI. This automation improves patients’ care access and also helps reduce payer/provider friction by providing intel that supports complex prior authorization processes for nonclinical SDOH services.

ML will help analyze data to build cohorts and make predictions, providing insights and visibility into the drivers of social risk. This enables the identification of high return on investment intervention opportunities, such as food literacy, economic stability, transportation, and housing, ensuring a person’s engagement within their care journey is not interrupted.

GenAI, on the other hand, will analyze data and offer a new context of care with deeper clarity and granularity, leading to more positive outcomes. It is powerful in the sense that it can do many things simultaneously.

GenAI’s ability to consume and analyze complex data from different sources to predict emerging risks is vital. It can also determine which care pathways would yield the most positive outcome for an individual or population cohort, which is a game changer in U.S. healthcare. GenAI is a powerful technology for addressing SDOH needs as well as other emerging risks, ultimately supporting the value-based care model by improving member satisfaction, lowering costs, enhancing the quality of care, and reducing care team burnout.

About Pravin Pant

As Vice President of Advanced Analytics, Pravin leads ZeOmega’s advanced analytics team, working on new and existing AI solutions and social determinants of health (SDOH) solutions with a goal of creating best-in-class software products. Pravin is an experienced executive leader with a demonstrated history of successfully working in the hospital & health plan industry for over 20 years. He is deeply proficient in team building, leadership, innovation, artificial intelligence, big data, predictive analytics, and population health management.

The following is a guest article by Pravin Pant, Vice President, Advanced Analytics at ZeOmega

As much as 80% of an individual’s health is determined by social and economic conditions. Social determinants of health (SDOH) are pivotal in shaping population health, providing care teams with a comprehensive understanding of the factors affecting their patients.

Health plans and providers recognize that challenges such as food insecurity or inadequate housing hinder optimal health outcomes without additional support extending beyond clinical care. Therefore, healthcare organizations must implement effective strategies to gather SDOH data and apply targeted interventions to improve population health outcomes. By adopting the right approaches, organizations can address SDOH and ensure patients receive the care they need, while simultaneously reducing costs, emergency services utilization, and employee burnout.

To build effective SDOH strategies, payers and providers must leverage technology such as artificial intelligence (predictive and GenAI) to integrate diverse data from both public and private sources, including assessments, surveys, and social media. It is important to collect community-level SDOH information from publicly available databases, mapped to the census tract level, and combine it with experiential claims data and personal assessments to create comprehensive digital member profiles. These profiles are critical for machine learning (ML) and predictive next steps.

When health information is collected from a variety of databases, it becomes invaluable to care teams, providing visibility into SDOH risks alongside behavioral and clinical risks within a population cohort or an individual. Insights may identify the probability of homelessness, possible chronic conditions, and other risks such as opioid overdose. The information, when made available at the point of care, allows clinicians to make proper care assessments and transition of care. Addressing risks early enhances the quality of care, while unaddressed risks can lead to a decline in overall care quality.

By leveraging GenAI, care teams can adopt a holistic, whole-person care approach that addresses medical, social, and behavioral health. GenAI also guides the next best preventive care strategies and their implementation for improved outcomes. But generally, the main idea is that GenAI can help alert care teams to multiple health issues that can be addressed and prevented simultaneously. From the health plan perspective, GenAI helps care teams tailor interventions for current and predicted issues. For example, research studies and data show that people at high risk for education and transportation have much higher poverty levels.

In addition, by identifying high SDOH risk members early, healthcare organizations can be proactive in addressing the non-clinical/social needs so that when/if a member develops clinical risk such as diabetes, both the member and plan will be better able to manage the challenge. However, it is important to remember that predictions are only as good as the data they are based on. Therefore, having a sound interoperability strategy in place coupled with an SDOH strategy is critical. Interoperability plays a major role in successfully managing and exchanging data between systems, which in turn supports the generation of effective predictive models that health plans need to develop and deploy strong intervention approaches.

SDOH strategies and solutions can be enhanced with a predictive next step, meaning clinical users can include the comprehensive member profile generated from EMR information and member behaviors. By applying ML methodologies, individual members’ or cohorts’ needs can be met, helping clinical organizations be proactive about addressing the nonclinical/social needs through automated next best action and intelligent care workflows. Because these insights are more precise, they might suggest new kinds of health initiatives that may have been historically overlooked or never considered (e.g., sending a mobile clinic to certain neighborhoods where transportation options are limited or starting a “how to cook” program in areas that face food insecurity).

GenAI, such as GPT-4 framework, can be critical when it comes to understanding individuals’ circumstances their doctors and health plans may not be aware of. It can help provide visibility into a person’s SDOH needs that were unable to be identified during direct interactions and recommend the precise downstream action for a member or cohort. Furthermore, GenAI frameworks help validate and justify why a person will need certain SDOH help and auto-approval. For example, if a physician needs to send a prior authorization for an MRI, the system will also flag that the patient will need transportation to complete the scheduled MRI. This automation improves patients’ care access and also helps reduce payer/provider friction by providing intel that supports complex prior authorization processes for nonclinical SDOH services.

ML will help analyze data to build cohorts and make predictions, providing insights and visibility into the drivers of social risk. This enables the identification of high return on investment intervention opportunities, such as food literacy, economic stability, transportation, and housing, ensuring a person’s engagement within their care journey is not interrupted.

GenAI, on the other hand, will analyze data and offer a new context of care with deeper clarity and granularity, leading to more positive outcomes. It is powerful in the sense that it can do many things simultaneously.

GenAI’s ability to consume and analyze complex data from different sources to predict emerging risks is vital. It can also determine which care pathways would yield the most positive outcome for an individual or population cohort, which is a game changer in U.S. healthcare. GenAI is a powerful technology for addressing SDOH needs as well as other emerging risks, ultimately supporting the value-based care model by improving member satisfaction, lowering costs, enhancing the quality of care, and reducing care team burnout.

About Pravin Pant

As Vice President of Advanced Analytics, Pravin leads ZeOmega’s advanced analytics team, working on new and existing AI solutions and social determinants of health (SDOH) solutions with a goal of creating best-in-class software products. Pravin is an experienced executive leader with a demonstrated history of successfully working in the hospital & health plan industry for over 20 years. He is deeply proficient in team building, leadership, innovation, artificial intelligence, big data, predictive analytics, and population health management.