Digital health initiatives have reshaped how healthcare organizations deliver care, manage operations, and engage patients — but not always in the ways leaders anticipated. From patient portals and EHR implementations to AI-powered automation and remote monitoring, the industry has made significant investments in technology that promised to transform care delivery. The results, many say, have been uneven at best.
Becker’s asked 35 healthcare executives and clinicians from across the country to reflect on the digital initiatives that underdelivered at their organizations — and what they learned from the experience. Their responses revealed a consistent theme: technology rarely fails on its own. Change management, clinician trust, workflow integration and organizational readiness are often the difference between a tool that transforms care and one that collects dust.
The leaders featured here are speaking at Becker’s 11th Annual Health IT + Digital Health + RCM Conference, set for Sept. 14-17 at the Hilton Chicago.
If you would like to join the event as a speaker, please contact Scott King at sking@beckershealthcare.com.
As part of an ongoing series, Becker’s is connecting with healthcare leaders who will speak at the event to get their perspectives on key issues in the industry.
Editor’s note: Responses have been lightly edited for clarity and length.
Question: Which digital initiative underdelivered — and why?
Sandhya Chandrasekhar. Associate Vice President, Enterprise Analytics and Data Engineering for Memorial Hermann Health System (Houston): We measured deployment and usage. We should have measured decisions.
Every dashboard and analytics product we built had users, not always owners. We were tracking logins and usage, not outcomes. We used to think our job was done at delivery. Shifting our strategy to track decision velocity, how fast does an insight actually move to an action: changed how we build, how we measure, and how we define “done”.
Delivered isn’t done. Decided is.
Danny Lee, MD. Chief Medical Informatics Officer for Johns Hopkins Medicine (Baltimore): Autonomous AI tools in the EHR have broadly underdelivered relative to their promise. AI-assisted patient messaging and automatic inbox routing require so much provider oversight that net time savings are marginal, and electronic prior authorization still demands significant manual intervention despite years of being positioned as a solved problem. The contrast with ambient AI documentation is striking. That category has delivered real, measurable time savings and continues to expand rapidly. The difference comes down to workflow fit; ambient documentation augments what physicians do naturally, while inbox and prior auth automation asks clinicians to supervise processes that are not yet reliable enough to trust.
Judd Hollander, MD. Senior Vice President of Healthcare Delivery Innovation and Chief Virtual Care Officer for Jefferson Health (Philadelphia): I think remote patient monitoring has underdelivered for several reasons. Unlike new pharmaceutical discovery process where clinical trials are critical before utilization of the drugs, RPM device companies generally do not perform large scale clinical trials to validate impact on meaningful clinical outcomes. The cost for these devices, the lack of clinical evidence demonstrating long term value, and the poor reimbursement model make this very challenging to implement at scale. I would recommend that when start-ups in the digital health space are doing their financial raise, they include money for clinical trials to prove the impact of their devices. This can then drive reimbursement models that can help drive adoption.
George Bailey. Director, CyberTAP, Technical Assistance Program for Purdue University (West Lafayette, Ind.): If you look back at CMS Meaningful Use, the big digital initiative that really underdelivered was true nationwide EHR interoperability. Meaningful use succeeded at getting hospitals and doctors to adopt EHRs, but it mostly rewarded checking boxes — like sending documents — rather than making data flow easily in everyday clinical care. Vendors and providers still had incentives to keep systems closed or customized, and early rules didn’t strongly enforce anti–data blocking. The result was a lot of digital records, but far less real, usable data sharing than policymakers and clinicians expected. My family of four has 9 different patient portal accounts, not to mention laboratory, imaging or urgent care notes … not very meaningful from a patient’s perspective.
John W. Gachago. Vice President, Digital Innovation for Parrish Healthcare (Titusville, Fla.): Many organizations invested early in predictive models, virtual assistants, or automation, but underestimated the readiness of their data, governance, and clinical or operational workflows. As a result, pilots showed promise, but production impact lagged due to fragmented data, limited clinician trust, and insufficient change management. Technology moved faster than organizational adoption, turning innovation into isolated tools rather than sustained, value‑driven transformation.
AI adoption improves when organizations treat it less as a technology rollout and more as a workforce and workflow transformation. Practically, that means anchoring AI use cases to real operational pain points, involving clinicians and frontline staff early in design, and setting clear accountability for outcomes — not just pilots.
However, change management is the multiplier. Leaders must invest in trust‑building (transparency about how models work and their limits), capability building (role‑based AI training, not one‑size‑fits‑all), and incentive alignment so teams are rewarded for adoption and improvement. When AI is embedded into daily workflows, supported by strong governance and continuous feedback loops, it shifts from “experiments on the side” to a durable driver of clinical, financial, and access outcomes.
Charleen Singh, PhD. Program Director, DNP-FNP Program for University of California, Davis Betty Irene Moore School of Nursing (Sacramento, Calif.): In the complex world of patient assignments in the acute care setting the digital initiatives around staffing have underdelivered. The why is likely answered by the realization that nursing underestimated the work, tasks, and time for critical thinking that is required to safely care for patients when digital staffing models were built. To further challenge the delivery of digitalizing staffing models and staffing matrixes was the unexpected increase in patient complexity and changes in documentation requirements.
Pete D’Addio. Chief Technology Officer and Vice President, Technology for LCMC Health (New Orleans): At my organization, one initiative that underdelivered was a clinical communication platform. Despite thorough requirements gathering, workflow validation and a solid ROI model going in, the realized outcomes fell short for our clinical staff and created more challenges than value. We are currently evaluating alternative platforms better suited to our clinical workflows. The key lesson is that pivoting is costly in both time and dollars, but the long-term ROI of getting to the right solution outweighs the short-term pain of changing course.
Priya Kumar, MD. Vice President, Medical Affairs and Chief Medical Officer for Self Regional Healthcare (Greenwood, S.C.): In a rural health system, the patient portal was a humbling reminder that technology alone does not equal transformation. Despite being a longstanding part of modern healthcare, many of the same barriers still limit its efficient use—added administrative burden for clinicians and staff through heavier inbox volume, workflow friction, and uneven adoption in communities where broadband access, digital literacy, and trust in digital communication remain real challenges. It taught us that digital tools do not become effective simply by existing; they require operational redesign and an honest understanding of the digital maturity of the patients we serve. That lesson has shaped how we approach every digital initiative since.
Sam Afirmar, MD. CMO and Chief Information Officer for The Brooklyn Hospital Center (New York City): I think the blockchain hype from a few years ago is a great example of a digital initiative that underdelivered. It was touted as a transformative tool, borrowed from cryptocurrency, that could be seamlessly applied to medicine. In reality, it never found a meaningful place in healthcare. It felt like a blockchain hammer looking for a nail to pound. While there are narrow use cases where it adds value, people started using the term without understanding the underlying technology, largely to seem like they were in the know. It became a proxy for following hype rather than genuinely understanding technology and healthcare problems.
Tony Sillemon, PsyD. Director, Community Health for Alta Bates Summit Medical Center, Sutter Health (Berkeley, Calif.): One digital initiative that has underdelivered across the industry is referral and care coordination tools within the EHR. While these platforms are designed to improve access and continuity, their impact is often limited when implementation does not fully align with clinical workflows or extend to community-based partners. The technology itself is strong, but success depends on thoughtful integration, consistent training, and cross-system collaboration. Organizations that invest in aligning people, processes, and technology tend to see the greatest return.
BABA Olanloye. IT Director for Pioneers Medical Center (Meeker, Colo.): One digital initiative that often underdelivers is technology implemented without sufficient workflow redesign or frontline adoption planning. Even strong platforms can fall short when organizations focus on the software itself rather than how people will use it day to day. Success usually depends less on the tool and more on leadership alignment, training, accountability, and measurable outcomes. In healthcare, digital transformation works best when technology is paired with operational change management.
Gerrit von Wenckstern. Marketing Director for The University of Texas Medical Branch (Galveston, Texas): An initiative that didn’t meet expectations was ER wait-time promotion. Transparency was the goal, but emergency medicine doesn’t function like retail; patients are treated by acuity, not by the clock. That disconnect made the data more directional than actionable and highlighted how easily good digital intentions can create unintended consequences. It underscored the responsibility healthcare organizations have to pair digital tools with education, particularly around appropriate site-of-care decisions which is something we continue to prioritize at UTMB Health.
Nipa Shah, MD. Chair of the Family Medicine Department for University of Florida Health (Gainesville, Fla.): One digital initiative that underdelivered was the rollout of AI and analytics tools that were not explicitly transparent to clinicians.
Many tools did not clearly explain how recommendations were made or how data was used/stored/deleted, which created confusion and mistrust.
In primary care and especially academic health systems like mine, this lack of transparency slowed adoption and delayed implementation. When clinicians do not understand or trust a digital tool, it is unlikely to deliver real value.
Bradley Locke, DO. Chief Medical Information Officer for Prevea Health (Green Bay, Wis.): In my opinion, one of the largest digital initiatives that underdelivered was the “Meaningful Use” program. This program drove rapid adoption of Electronic Health Records but often underdelivered on true care transformation. Much of the effort focused on meeting prescriptive reporting measures rather than improving clinical workflows or outcomes. This led to checkbox-driven documentation, increased administrative burden and limited gains in interoperability despite technical compliance. As a result, organizations achieved certification but not the expected improvements in efficiency, clinician satisfaction or patient outcomes.
Enitza George, MD. Associate Professor and Chief Population Health Officer for SUNY Downstate Health Sciences University (New York City): A digital dashboard initiative underdelivered because it prioritized visibility over action. We were able to move metrics from red to yellow, but the underlying care processes didn’t fundamentally change. Dashboards are not interventions; they are tools to support redesigned workflows. Without accountability structures, incentives and operational ownership, metric improvements can create the illusion of progress without true system transformation.
JohnRich Levine, DNP. Chief Nursing Officer for Reeves Regional Health (Pecos, Texas): One digital initiative that often underdelivers is the patient portal. Many organizations install the platform, then stop short of redesigning the work around it. Patients want speed, clarity, access to scheduling, refill requests, results and simple communication. Staff want fewer calls and less rework. When the portal helps both groups, adoption grows; when it adds steps, usage stalls.
Sheri Strobel. CIO for Chapters Health System (Temple Terrace, Fla.): One area that consistently underdelivers is digital investments aimed at saving small increments of time — like 15 to 20 minutes per day — across large teams. While those gains sound meaningful, they rarely translate into measurable financial ROI because staffing models and workflows remain unchanged. The gap isn’t the technology itself; it’s the lack of end-to-end workflow transformation. When organizations automate isolated tasks without redesigning the full body of work, the savings stay theoretical instead of operational. Real impact comes when digital initiatives are paired with intentional workflow redesign and staffing alignment to capture value at scale.
Chuck Christian. Vice President, Technology and Chief Technology Officer for Franciscan Health (Mishawaka, Ind.): There are several digital initiatives that I’ve been involved with during my career, unfortunately, the ones that were less than successful were rarely due to the ability of the technology, but rather due the lack of adoption and leadership support and/or cultural rough pavement. Another reason that I’ve experienced initiative failing was due to them looking like or being perceived as IT initiatives. I’m also thought and tried to promote any initiative as operationally lead and technically supported, especially in the clinical areas.
Bob Berbeco. CIO for Mahaska Health (Oskaloosa, Iowa): The digital initiative that underdelivered for us was not a single solution as much as an approach. The approach underdelivered because it emphasized technology more than the hard work of operationalizing a solution to create durable value for clinicians, physicians and the organization. We were evaluating isolated, bolt-on third-party, AI-based solutions that were not going to be tightly embedded in our clinical workflows and, in some cases, were likely to be duplicated by our Epic and Community Connect roadmap.
We regrouped and established a data science and AI center of excellence, a tiered risk-reward matrix-based intake process, and a workflow-native strategy focused on optimizing physician and clinician time while scaling native Epic capabilities where they can create the most sustainable value. We are also building an AI portfolio in house that is practical, governed and value-driven, with near-term focus on revenue cycle opportunities such as claims analysis, payer configuration, denial reduction and underpayment recovery, while also advancing clinical documentation support and operational analytics where AI can return time and improve decision-making.
The lesson learned is that sustainable innovation is not measured by how advanced the technology is; it is measured by whether it becomes part of the work, improves the experience of the people doing the work, and continues to create value after the initial excitement fades.
Beth Carlson. Chief Revenue Officer for The Ohio State University Wexner Medical Center (Columbus, Ohio): I think many will point to prior authorization automation as an underwhelming digital initiative, which actually exposes the real issue. The constraint isn’t the technology itself; it’s the environment that it operates in. Payer variability, clinical nuance and fragmented external systems create structural limits that automation alone can’t overcome. Digitization falls short when the data and decision rules it depends on are fragmented across inconsistent systems. In that environment, even the most advanced capabilities will eventually hit a ceiling, which is why most efforts today still optimize discrete steps with a process rather than integrating flow across them. To truly move the system forward with the patient at the center, digitization has to extend across payer data, evolving medical policies, contractual frameworks, and state-level systems. The opportunity is a shift from isolated automation to integrated intelligence.
Brian Lancaster. Senior Vice President and CIO for Children’s Mercy (Kansas City, Mo.): The effort that fell short was physician secure communications. Adoption reached 88%, but communications require 100% participation by all physicians and medical departments. Buy in and change management were not achieved. Less than full adoption in a requirement of full adoption is like trying to cross a 10-foot chasm in two 5-foot jumps, it fails. The lesson is that outcomes depend more on change management than on technology.
Chrisanne Timpe, MD. Hospitalist and Clinical Lead, HealthPartners Hospital at Home for HealthPartners Care Group (Bloomington, Minn.): It seems that we still haven’t hit the mark on initiatives to integrate patient-centered biometric data from remote locations seamlessly into the EMR in a way that is accessible to clinicians and affordable to patients. While plenty of tools exist, leaders in care delivery are left with the challenge of considering form factors, EMR integration challenges and clinical applicability, while under tremendous financial pressure. This continues to be a challenge in the Hospital@Home space.
Ruchi Garg, MD. CMO and Gynecologic Oncologist for Fairview Park Hospital (Dublin, Ga.): One digital initiative that has consistently underdelivered is large-scale EMR implementation, especially when it’s been driven more by regulatory requirements than clinical usability.
The intent was to standardize care and improve outcomes, but in practice it’s created a lot of what I call system bloat. By that, I mean too many clicks, redundant data entry and long notes that don’t add much clinical value but take a lot of time to complete.
Two issues really stand out. One is the lack of interoperability, so clinicians are often piecing together fragmented information instead of seeing a complete patient picture. The other is how regulatory requirements have expanded documentation well beyond what’s useful in clinical care.
The impact has been real. Clinicians are spending more time in the EMR and less time with patients, which contributes to burnout and takes away from the patient experience. In many ways, the technology that was supposed to support care has ended up getting in the way.
The good news is that this shift is starting to be recognized. There is growing alignment around the need to reduce documentation burden and improve interoperability, and we are seeing AI-based solutions being developed and deployed to help connect data more seamlessly and take some of that load off clinicians.
Sunil Dadlani. Executive Vice President, Chief Information and Digital Transformation Officer and Chief Cyber Security Officer for Atlantic Health (Morristown, N.J.): Digital front door is a great example of an initiative many health systems have struggled to fully optimize or implement effectively. It often underdelivered because organizations digitized the entry point without redesigning the access engine behind it. Patients could search, click, and request care more easily, but the system still lacked the right capacity, scheduling rules, governance and operating discipline.
Joy Oh. Chief Information and Digital Transformation Officer for Christ Hospital Health Network (Cincinnati): AI-drafted messages have not been highly adopted by our clinicians for three primary reasons. Accuracy was one issue, with some responses including incorrect advice or medication errors. The AI also struggled to handle patient messages that covered multiple topics and questions in one note, like scheduling, billing and clinical updates. Most importantly, many messages missed the mark on tone — either sounding too clinical, too wordy or not reflective of how clinicians naturally communicate with patients. While AI has come a long way, there’s still progress to be made in matching the empathy, judgment and nuance that are essential in complete and compassionate patient care.
Sandeep Kashyap, MD. Surgical Director of Lung Cancer Screening and Thoracic and Esophageal Surgeon for CAMC Thoracic and Esophageal Surgery at Vandalia Health (Charleston, W.Va.): As a thoracic surgeon, I use the most advanced surgical robotic platforms daily for complex lung resections, esophageal surgery, and mediastinal procedures.
In my experience, however, the digital initiatives meant to support those systems and the providers perioperatively are often under delivered. On paper, layering AI-driven tools onto the OR workflow — for real-time guidance, ambient listening, EHR integration and supply chain tracking — sounds transformative.
In practice, these systems lag way behind the precision of the robots themselves. We deal with constant manual overrides and clunky interfaces that force us to spend more time clicking through fields, dealing with lack of seamless communication amongst team members, chasing missing trays or placing equipment orders than operating. The result is widespread burnout among surgeons, nurses, and residents who feel like they’re practicing medicine on a computer rather than at the bedside.
The biggest barrier to real AI progress is the persistent gap between IT developers and actual clinical workflows. Too often, these initiatives are pushed from the top down without enough input from the surgeons and healthcare workers who use them every day. Engineers often lack real insight into the realities of the OR, and the cost of custom integration makes adoption prohibitively expensive. Yet the true value of these tools extends far beyond a three-year ROI. It’s about preserving our irreplaceable workforce and unlocking human potential so we can focus on patient care instead of wrestling with technology.
I remain convinced AI will augment surgery far more than it does now—particularly in clinical decision making, intraoperative image processing and interpretation, precision medicine, and perioperative workflows—once we close the gap on seamless integration and true real-time performance. It has the potential to transform preoperative planning and risk stratification.
But most importantly, it will not replace the surgeon — or any essential healthcare worker. The technical skill and ongoing judgment required to perform safe surgical operations simply cannot be replaced. The human elements of compassion, touch, and empathy simply cannot be automated.
Whether it’s sitting with a family to gently deliver a cancer diagnosis, holding a hand when we’ve successfully cured the disease, or standing with them in quiet grief when things don’t go as planned, those are the moments that define our work and we need to strengthen this perspective.
Romila Aloysius. Assistant Vice President, Heart and Vascular Institute for AdventHealth (Altamonte Springs, Fla.): In my experience leading digital health strategy and product development, I’ve learned initiatives underdeliver when they are deployed as technologies rather than designed as scalable operating models across the full product lifecycle. Digital front doors broadened access but often fell short in advancing longitudinal coordination across the patient’s journey. Similarly, many analytics investments allowed data maturity to outpace management maturity, generating insight without the governance and decision structures needed to act on it. AI, automation and remote monitoring have faced similar challenges when layered onto siloed workflows or launched as technology pilots without the care models, reimbursement models, and operational redesign required for scale. The lesson is consistent: digital creates value only when embedded in system strategy, not treated as a stand-alone innovation.
Diane Constantine, MSN, RN. Senior Director, Enterprise Health Informatics for Children’s Hospital of Philadelphia: One area that underdelivered was our early approach to Epic Gold Stars utilization work. We had strong configuration and clear opportunity, but the impact didn’t fully materialize because we approached it too much from a build and metric perspective, rather than embedding it into clinical workflows and existing operational governance. We also hadn’t yet fully leveraged our service line workgroups to create ownership and accountability, and the connection to frontline priorities wasn’t always clear. As a result, engagement and adoption were more variable than expected, which is what led us to evolve the model to be more integrated and operationally driven.
Aron Klein. Vice President, Finance Operations and Supply Chain for Carle Health (Urbana, Ill.): Carle Health has adopted several robotic process automation solutions to assist our team members and the organization by creating efficiencies in workflows and completing repetitive tasks quickly and efficiently. However, RPA has its limitations, and we’ve found several situations where RPA just won’t solve the problems we’re trying to address. For example, RPA may break easily due to minor payer website changes creating downtime and manual processes for team members. In addition, we’ve been working with a vendor for nearly two years to automate the response to payer documentation requests and have just not been successful in creating a solution. Ultimately, we believe RPA has its use cases, but we need to move to more advanced AI solutions in many cases.
Yusuf Sermet, PhD. Associate Professor, ByWater Institute and Adjunct Faculty, Pediatrics, School of Medicine for Tulane University (New Orleans): Early digital initiatives that relied solely on generic, cloud-based probabilistic AI models have often underdelivered in clinical settings because they function as unexplainable black boxes that clash with the rigid privacy and operational demands of healthcare. The future of clinical AI lies in hybrid solutions that combine the raw capability of advanced models with deterministic rules, clinical checks and balances, and strict operational boundaries. This means keeping data strictly localized. While this can involve dedicated on-premise GPU clusters, we can now customize highly optimized models to run locally on standard $3,000 to $4,000 workstations, distributing the computational weight across existing office hardware. This localized, sovereign infrastructure ensures absolute patient privacy and ultralow latency, while shielding hospitals from the fluctuating costs and shifting quality of third-party cloud AI services. By becoming technologically self-reliant, health systems can budget predictably and capture the immense benefits of next-generation AI, all while keeping the ultimate decision-making responsibility squarely in human hands.
Nicole Benitez. Director of Transplant Quality, Compliance and Data Analytics for Emory Healthcare (Atlanta): The biggest underperformer in my world has been the suite of patient-facing digital tools (portals, follow-up apps, remote monitoring) pitched as a path to patient activation and balanced clinical work. Adoption tracks closely with health literacy, language and broadband access, which means the patients who most need post-discharge engagement are the ones least likely to use the tool. In transplant, where the post-discharge window carries real graft and life consequences, that gap shows up as missed labs, medication non-adherence and avoidable readmissions. Digital deployment is not a substitute for human follow-up. Until we staff for the patients the tools miss, the ROI in this category will keep underdelivering.
Philip Bernard, MD. Senior Vice President and Chief Medical Information Officer for Children’s Health (Dallas): There are many – most fail due to hurdles with workflows, rather than the technology itself. One initiative that comes to mind was an Alexa-type device for inpatients where the nurses were required to obtain separate consent. It added just one more thing for the nurses to do — ultimately the project failed from lack of use. Another example is rural ED to quaternary ED video consults. Both ED physicians are incentivized to transfer the patient out of the primary ED as soon as possible. Keeping the patient in the rural ED added cognitive load for both the receiving and sending physicians. These initiatives also fail from lack of use.
Trying to understand what problem we are trying to solve — making the right thing to do built into workflows — brings ROI. Technology uses designed around the technology seldom bring results.
Alexander Levit, MD. Medical Director, Hospital at Home for Lee Health System (Fort Myers, Fla.): A digital initiative that, to date, has underdelivered would be the combined attempts at adding price transparency into digital charts. This has not panned out at the physician level because, I believe, physicians have yet to be offered any serious way to know what the patient will pay for a test or treatment. Major initiatives so far have largely included qualitative dollar sign symbols. When quantitative prices were actually listed, they were the low prices reflecting Medicare’s per-lab reimbursement rates, which are often not reflective of a particular patient’s context.
Dave Newman, MD. Chief Medical Officer, Virtual Care for Sanford Health (Sioux Falls, S.D.): One digital initiative that has not yet fully delivered is the surge of direct-to-consumer digital mental health apps. The evidence shows a consistent pattern: most lack rigorous proof of efficacy, and while initial engagement can be strong, adherence drops off quickly — often within the first two weeks. That highlights a fundamental gap. Mental health care isn’t episodic or self-directed; it requires clinical guidance, continuity and accountability — elements many standalone apps aren’t designed to provide.
In rural communities, this challenge is even more pronounced. Access is a real barrier, and digital tools create meaningful opportunities to bridge that gap — but outcomes still depend on connection to trusted providers and coordinated care over time. What stands out is the need to build on what these tools get right — convenience and reach — and embed them into provider-led models like telepsychiatry and integrated behavioral health. That’s where digital moves from short-term engagement to sustained, measurable impact.
Rosy Thachil, MD. Director of the Cardiac Intensive Care Unit and Cardiology Division for Elmhurst Hospital Center and Assistant Professor of Medicine for Mount Sinai College of Medicine (New York City): A common pitfall in digital health is overestimating technology and underinvesting in change management. Even well-designed tools can underdeliver if they aren’t supported by training, culture shift, and operational alignment. In complex environments like the ICU where I round, where decisions are time-sensitive and high-stakes, digital solutions must enhance — not disrupt — clinical intuition and teamwork. The most successful initiatives I’ve seen are those that are iterative, clinician-informed and tied to improvements in care delivery. Technology alone doesn’t transform care — people and systems do.
Toyosi Olutade, MD. CMOfor UnityPoint Health-Quad Cities (Bettendorf, Iowa): We’ve invested heavily in EHRs, but they still haven’t delivered the seamless, connected system we hoped for — many teams are still relying on workarounds like faxing. A big part of the issue is the lack of interoperability between platforms, along with systems that weren’t built for real-time data exchange, leaving teams to manually move information across settings. In many ways, EHRs have become expensive data repositories that are significantly underutilized. On top of that, complex payment models and regulatory requirements make it difficult to scale digital solutions beyond local environments. Too often, change management and clinical input are underemphasized, and these efforts are treated as IT projects rather than the care delivery transformation they really are.
Karen Walker. Enterprise Director, Cardiovascular Services for Santa Clara Valley Hospitals (San Jose, Calif.): The digital initiative that underdelivered was our early effort to move from traditional data extraction of data to more real-time, AI-enabled operational model. The initiative underdelivered not because AI lacked value, but because the organization had not yet built the readiness, governance, and workflow discipline needed to turn digital capability into measurable performance. That experience shaped how I now think about AI adoption: start small, solve a real operational problem, measure adoption, and scale only after the workflow proves value.
Parag Jain. Director, Clinical Research for Children’s Health and University of Texas Southwestern Medical School (Dallas): Many clinical AI initiatives have underdelivered — not due to lack of data or model sophistication, but because they misunderstand clinical workflow and how clinicians think under pressure. Most tools generate predictions without embedding the prioritization and troubleshooting frameworks that drive real-time decision-making at the bedside. This disconnect often increases cognitive burden rather than reducing it. Until digital health tools are designed to mirror clinical reasoning — not just statistical accuracy — their impact will remain limited.
The post Why digital health initiatives fail: 35 healthcare leaders weigh in appeared first on Becker's Hospital Review | Healthcare News & Analysis.
