Mapping Markets
September 18, 2024

Clinical Pathways Market Map: How AI is shaping the development of clinical pathways

Patrick Wingo's headshot
Patrick Wingo
Head of Research, Elion

This is part of Elions weekly market map series where we break down critical vendor categories and the key players in them. For more, become a member and sign up for our email here.

Given the rate of progress in medicine, it’s become an impossible task for clinicians to map out every guideline and treatment protocol. At the same time, it’s critical that every patient receives consistent, evidence-based care across medical settings.

Classically, this problem was solved by mountains of notecards, a solution that was neither user-friendly nor scalable. Over time, these paper reminders evolved into clinical pathways software integrated into the EHR. Today, we’re seeing even more utility with AI-enabled pathway products that can offer a more dynamic approach to implementing evidence-based medicine.

When Standardized Care Makes a Difference

At its core, evidence-based medicine plays a critical role in standardizing treatment protocols for specific conditions, ensuring consistency across providers and improving patient outcomes. Deviating from best practices or missing a critical intervention at the right time can lead to poorer outcomes, delayed recovery, or even preventable complications.

Clinical pathways are crucial for:

  • Managing chronic conditions: Ensuring evidence-based care for diseases like diabetes, heart failure, and COPD.

  • Oncology care: Standardizing cancer treatment plans to ensure consistency and that the latest evidence is applied.

  • Post-operative recovery: Outlining timelines and interventions for post-surgical patients to ensure faster and safer recovery.

  • Emergency care: Providing clear protocols for managing strokes, heart attacks, or trauma.

  • Quality improvement: Aligning care with hospital performance metrics and reducing unnecessary variations in care.

Bridging the Gap Between Guidelines and Practice

However, the development and implementation of these pathways is far from straightforward. Clinical pathway developers must sift through vast volumes of clinical research, expert opinions, and regulatory guidelines, converting these guidelines and protocols into discrete logical flows that can be implemented in the EHR. Traditionally, as guidelines needed to be updated, developers had to rework the logic powering the workflows.

On the provider side, implementing new clinical pathways can be equally challenging, as they require changes to workflows and coordinated care across large teams. Further, using care pathways still requires a great deal of judgment, as objective patient data has to be pulled from across medical records, and subjective patient data has to be interpreted against guidelines. The complexity of care delivery is heightened in hospitals, where pathways need to be adjusted against individual patient needs and comorbidities without losing sight of standardized care objectives.

While clinical pathways provide consistency, they also require careful management:

  • Updating guidelines: Ensuring pathways reflect the latest medical research and evolving standards.

  • Customizing for patients: Adapting pathways to individual patient profiles without compromising evidence-based care.

  • Coordinating across teams: Ensuring multidisciplinary teams adhere to a shared care plan, minimizing miscommunication.

Technological Advancements in Clinical Pathways

GenAI is reshaping how clinical pathways are developed, managed, and applied. Instead of hand-coding every logical rule, agents are now capable of interpreting guidelines through retrieval-augmented generation (RAG) and other related techniques, reading in both subjective and objective data from EHRs, and providing more tailored guidance to clinicians. While traditional clinical pathways encoded boolean logic with well-defined variables, GenAI allows the combination of fuzzy logic to interpret complex data in conjunction with well-defined procedures and guidelines.

What makes this hard to execute correctly, though, is the fact that hallucinations can regularly occur, especially as the RAG datasets increase in size and the prompts get more complex. As a result, the technological challenge is in managing hallucinations and ensuring the clinical pathways and underlying data are accurate.

Differentiating Clinical Pathways Vendors

This subset of clinical decision support vendors is focused across a number of different use cases, but can be differentiated by their EHR integrations, depth of those integrations, and their support for AI-enabled workflows. Some of the legacy CDS systems may not have AI yet, but they are likely moving fast to incorporate it into how they serve clinical guidelines.

  • Products that have done the heavy lifting with integrating into health system EHRs include AgileMD, AvoMD, Curbside, and UpToDate (which may be considered more of a medical reference tool which can be accessed from the EHR).

  • Meanwhile, vendors like Pathway Medical have focused on building a mobile assistant to help clinicians query for the latest guidelines on their phones without EHR integration.

  • The final differentiation in this category, irrespective of modality, is specialty-specific clinical pathways solutions. Oncology-focused products like Flatiron Assist, EvidenceCare’s OncologyCare, Elsevier ClinicalPath, and Eviti Advisor are focused on oncology-specific workflows and EHRs due to the rapidly evolving and complex nature of oncologic evidence-based medicine. On the value-based care side of things, products like MedPearl and Orion Care Pathways are helping provide specialist insights to the front line of clinicians.

The Future of Clinical Pathways

A number of clinical pathways solutions have gone beyond clinical decision support, and have entered into clinical decision-making territory, with the clinician involved to sign off before treatment or orders are finalized. This transition raises several critical challenges for healthcare providers and the healthcare system as a whole.

From a regulatory perspective, we’re not well-equipped to handle it yet, as FDA guidance for software as a medical device (SaMD) tends to be for a specific diagnostic or treatment pathway, not for replacing a whole range of workflows. More importantly, medicine is a fundamentally human process that requires weighing the complexities of patient preferences, ethical issues, and nuanced clinical judgment, and we see most of the value here accruing in helping clinicians make more accurate decisions, rather than taking over their workflows.