Imagine being able to predict when a patient might need to see a specialist and then suggest the very best available and accessible specialist for that patient's unique situation. Using the power of AI, Conduce Health helps make this “good doctor for whom at the right time” possible for leading value-based care organizations today.
The Opportunity for Improvement
Let’s start with why these capabilities are needed. First, the prevalence of chronic conditions and the shortage of specialist physicians create supply/demand constraints that make care delivery far more complex for risk-bearing entities.
A primary example is the compounding effect of appointment wait time. As patients wait to see a specialist, the likelihood of no-show and leakage increases, their disease progresses, and costly ED and in-patient usage jumps 8.7 times higher.
More importantly, the absence of an established baseline for specialty care referral has given way to inefficient processes that generate suboptimal outcomes for all parties involved, including patients. With PCPs shouldering most of the administrative load, introducing technology that supports and amplifies a physician’s abilities can pave the way for greater efficiency and collaboration.
Conduce uses AI to make two distinct predictions:
- Need for Referral: Whether a patient requires a referral or visit to a specialist, and if so, in what time horizon.
- Personalized Specialist Match: Which set of specialists are optimally matched for specific patient cohorts based on their unique Patient Signature.
Providing insights around these two distinct predictions to value-based care organizations advances the “good doctor for whom at the right time” opportunity. They identify the patients who are most at risk of disease progression and higher costs and ensure that they are able to find the most appropriate specialist for their care.
Predicting the Need for a Referral
Utilizing a comprehensive Medicare population dataset, Conduce overlays multiple predictive models to assist healthcare organizations in identifying patients who may benefit from a specialist referral.
Individual precursors to project the near-term need and appropriateness of a specialty care referral include an increase in costs relative to an individual’s risk-adjusted benchmark, chronic disease stage progression, ED visits, or inpatient hospitalization for an acute exacerbation, etc.
Referral Need Prediction Highlights
- Conduce's predictive models utilize 250+ input features from the patient population including medical history, pace of disease progression, social determinants of health, recency of inpatient hospitalization, treatment information, and cost trends.
- Through continuous population analysis, these models not only recognize subtle patterns that indicate a patient’s condition is deteriorating, but they track indicators of hospital readmission risk and likelihood of a major procedure.
- Proactive identification of at-risk patients enables earlier intervention, potentially preventing serious health issues and reducing costs.
The goal is to use insights from the data to augment and support a physician’s ability to manage patients, giving them the tools to engage with the right patients and the right specialists at the right time to personalize care, slow disease progression, and potentially avoid costly and disruptive treatments.
Matching the Patient with the Right Specialist
After a patient is identified as needing specialty care, the next step is matching them to the most optimal specialist for their unique needs. To create this ideal match, we need to achieve a similar depth of understanding for the specialty care network available for that particular patient.
Much like a baseball player’s overall numbers can vary based on the conditions in which he plays, we find that the same is true for physicians. There are a multitude of factors indicating a specialist physician’s impact on the patients they see, which Conduce aims to help organizations better understand so they can offer more personalized care that is also in alignment with their value-based care.
Conduce aggregates and analyzes specialty care data based on a number of dimensions:
- Disease Type: The type(s) of conditions a physician manages has a huge impact on patient costs and outcomes. Conduce adjusts evaluations based on the mortality and severity of diseases treated (e.g., complex heart failure vs. routine hypertension management).
- Comorbidities: Patients with multiple health conditions present additional challenges and are more complex by nature. Models control for the complexity of these cases, ensuring physicians are credited with treating patients with more intense health needs.
- Age and Gender: Demographics are powerful determinants of healthcare outcomes. Elderly patients or those with specific gender-related conditions may require more intensive care.
- Social Determinants of Health (SDoH): Non-medical factors such as socioeconomic status, care accessibility, and living conditions play a significant role in patient outcomes. Conduce incorporates these factors to ensure physicians who serve disadvantaged populations are recognized.
- Practice Setting: The locations and affiliations of where care is delivered can significantly impact performance. Conduce accounts for optimal site-of-service for diagnostic tests and procedures, location where patients are seen, and the independence or health system affiliation of the provider.
Guided by Conduce’s patient cohorts, called Patient Signatures, personalized benchmarks can be created for each specialist based on their experience and past performance with each cohort. These benchmarks are then used to provide the most personalized match between patient and specialist. As an example, patients with heart failure with reduced ejection fraction (HFrEF) would be matched with cardiologists who excel in guideline-directed medical therapy (GDMT). Whereas patients diagnosed with heart failure with preserved ejection fraction (HFpEF) would be matched to cardiologists who have demonstrated significant experience managing several co-morbidities. Patients at earlier stages of disease would be matched with specialists who excel in creating preventive treatment plans. Patients further along in their disease would be matched with specialists who have demonstrated substantive experience in managing complex care.
Early results from Conduce’s data-driven approach to referral and network optimization have proven to be highly effective in helping risk-bearing entities deliver on the promise of high-value, affordable care. By leveraging data science, Conduce has made it possible for organizations to gain a deeper understanding of member populations and uncover the performance potential within specialty care to deliver personalized value-aligned care at scale.
To learn more about partnering with Conduce for value-based specialty care, please contact us today.
