La convergence de l'IA et de l'analytique des données

Technology   |   Andy Dé   |   Mar 4, 2020 TIME TO READ: 6 MINS
TIME TO READ: 6 MINS

A recent research survey, Data Analytics in Healthcare, by HIMSS Media, surveyed leaders employed in IT/cybersecurity, informatics, clinical, and business-related roles at U.S. hospitals and health systems and identified 4 key findings from the data:

  1. AI is a priority for many organizations, with 3 out of 4 respondents agreeing on specific areas for immediate implementation.
  2. Data analytics confidence decreases as complexity increases.
  3. Barriers to success with complex data analytics are multiple and diverse.
  4. The distribution of data analytics resources continues to tilt slightly toward clinical use cases at this time.

Despite that commitment to AI-infused analytics, only 56% of those surveyed rated their current descriptive analytics initiatives as extremely or very effective, and only 4 in 10 do so for predictive or prescriptive analytics. Clearly, it is a case of excitement tempered by a mix of caution and frustration.

Through conversations with practitioners and industry thought leaders at the recent AI4Healthcare Conference in NYC (articulated in this recent blog, Will AI augment hospital people, processes and systems, in the foreseeable future?), I’ve learned there is significant concern re: the lack of transparency manifested in a “Black Box approach” from many point solution vendors in the AI arena. Clinical leaders are rightfully apprehensive about leveraging the recommendations or acting upon analyses from algorithms that have not been validated by them or their experts, with life-impacting implications.

In addition to this “black box” anxiety, the research turned up other barriers to adoption, as illustrated in figure 1 below.

Figure 1. Primary barriers to the adoption of Artificial Intelligence (AI) in Healthcare today. Source: HIMSS Media

These barriers reflect a mismatch in the complexity of the tasks with the skill base and resources available to deploy AI-based solutions.

“The convergence of AI and analytics has drawn a lot of interest among healthcare systems ‒ more than 75% of respondents reported that AI will be of greater focus for the next year.”

‒ Ann Mackay, Insights Practice Leader at HIMSS Media

Healthcare Providers and Health Systems Plan to Invest in AI and Analytics in 2020-21

Despite the challenges, there is enormous excitement and enthusiasm around harnessing the promise and value from AI and analytics from a majority of the healthcare providers and health systems surveyed.

One of the key findings from the survey is the near-parity between the proportion of data analytics resources — as measured by budget, time, and/or talent — that are directed towards clinical use cases (e.g., population health and precision medicine) and back-office use cases (e.g., operations, supply chain management, HR, and finance).

Respondents said that about 53% of resources are applied to clinical use cases, while 47% are applied to the back-office ones. These findings are consistent with last year’s data, and respondents project the same distribution over the next 18 months.

This again is hardly surprising since clinical thought leaders have been “early adopters” of AI and Analytics to enable value with predictive and prescriptive analytics for population health segmentation and risk stratification (for acute and chronic conditions), and care coordination across the continuum of care (predicting utilization patterns and optimizing in-patient clinical throughput), where they perceive the highest value and impact from a patient and population health impact perspective.

Primary Areas of AI and Analytics Investment and Resource Allocation for Healthcare Organizations

The survey identified a prioritized ranking of use cases from an investment and resource allocation perspective, which is one of the most valuable insights from thisresearch for healthcare providers and health systems considering investments in AI and Analytics in 2020-21.

The top 3 areas of value from a clinical and population health perspective are:

  1. Proactively identifying patients at risk for adverse health events like heart failure and heart attacks.
  1. Predicting patient utilization patterns (e.g. missed appointments and optimizing inpatient/clinical throughput).
  1. Population health stratification including risk scoring for chronic diseases like cardiovascular disease, COPD, diabetes, and cancer.

Although clinical and population health management (PHM) are deemed a priority, the survey also reveals interest and awareness in other lines of businesses (LOBs), functions and use cases for AI investment and adoption, including:

  • Blending data from multiple EMRs, standardize curated data sets to secure a “single version of the truth” re: the patient.
  • Minimizing issues with reimbursement and payments including fraud, waste, abuse, and denials.
  • Enabling a 360-degree view of the patient for clinicians and nurses at the patient bedside.
  • Advancing precision medicine and personalized treatment of patients.
  • Improving supply chain management (SCM) efficiencies, including predicting stock outs and expired drugs.
  • Proactively identifying and mitigating cybersecurity risks, including medical fraud.
  • Identifying and classifying anomalies and imaging and incidental findings.
  • Proactively detecting fraud, waste and abuse pertaining to employee overtime and agency costs.

Given this level of awareness and prioritization, it is hardly surprising that 77% of those surveyed intend making significant investments in AI and analytics in 2020 -21, specifically targeted to enhancing their capabilities in the following areas:

  • Investing in more self-service analytics platforms, ideally those that offer AI capabilities.
  • Moving towards and standardizing on an integrated EMR platform.
  • Operationalizing analytics models to be leveraged by other departments and applications.
  • Training line of business (LOB) and end-users to up-skill their capabilities.
  • Hiring data scientists to create high value predictive and prescriptive analytics models.
  • Integrating AI and machine learning into their data analytics strategy given the convergence happening as we speak.

Information provided in this blog are results from the HIMSS Media Research Survey “Data Analytics in Healthcare”.

Key Takeaways

Leaders in healthcare see tremendous potential in AI and analytics to deliver on the promise of higher quality care at a lower cost by empowering their executives, business leaders, clinicians, and nurses by harnessing the power of predictive and prescriptive analytics.

Many healthcare organizations are seeking to harness the vast potential of AI to transform their clinical and business processes.

They seek to apply these advanced technologies to make sense of an ever-increasing “tsunami” of structured and unstructured data, and to automate iterative operations that previously required manual processing.

Tags