Anshuman Sharma

It was a pleasure to listen to Judd Antin, Dave Hora, and Christiana Lackner discuss “The Future of UX”  with Erin May and Carol Guest on User Interviews’ Awkward Silences podcast. One portion of the conversation focused on aligning user experience research with organizational aims and constraints. Such conversations appear to be occurring a lot in the UX field today, or at least they are amplified as of late. I find this surprising. When I transitioned into “industry” from academic industry I heard a lot about biases against academics due to a perceived lack of business acumen and understanding of organizational aims, constraints, and limitations. That biased perception never resonated with me. I have always felt that organizational and other structural considerations are part and parcel to research. In fact, due in great part to my direct experiences in the dumpster fire that is higher education today, I have an awareness of such things. Indeed, I would say that most academics are acutely cognizant of the implications of higher-level structures, as they were forced to survive and thrive amidst the stormy seas of higher education. Given this, I'm left somewhat puzzled and with an impression that many UX researchers lack, or at least underappreciate, awareness of at least some of the structural and organizational realities of capitalism that underlie most, if not all, of the work they do. If this diagnosis is correct, one suggestion is that researchers read Marx (and other social theorists, but maybe start with Marx). My first reading of Capital, Volume One, which I did with the accompaniment of wonderfully insightful lectures by David Harvey, was world-shifting. Marx’s work helped me to develop an initial understanding of the machinations of capital, which in turn led to greater understandings of the constraints, limitations, motives, and contradictions that drive regimes of capital accumulation. Beginning to understand and come to terms with the ultimate realities of those machinations–and the realities that those mechanics produce–is fundamental to the research that  I do today, no matter what the more proximal contexts are. Marx’s work also inspired me to adopt a dialectical approach to research and critical analysis. Dialectical thinking involves tacking back and forth between different scales and modes of analysis to obtain a more holistic view of structures, organizations, and processes, as well as their interrelationships. The approach allows me to recognize the interplay between various factors and develop a deeper understanding of the broader contexts in which my work is situated. By embracing such an approach, I can uncover more meaningful insights and produce research outcomes that have more nuance in terms of value propositions, strategy, and decision-making.

3

1 Comment

Introducing QAPS & i-QAPS : A New Approach to Monitoring Radiologist Productivity 👇 Productivity scoring systems are increasingly used in healthcare and radiology, but I find their current form deeply frustrating. Traditional scorecards focus almost entirely on volume—number of cases, type of cases, turnaround time—yet this only tells half the story. The most critical aspect of a radiologist’s work, the quality of their diagnoses, is often overlooked. That's why I want to open the conversation around a new concept: QAPS (Quality Adjusted Productivity Score). Errors in radiology are relatively common, with an error rate of 2-5%. These errors range from detection and characterization issues, such as missing significant lesions like tumors or aneurysms, or mischaracterizing a tumor as benign, to simpler documentation mistakes (like interchanging left and right sides). Fatigue and a high volume of cases increase the likelihood of these errors, especially toward the end of a shift. The problem with volume-driven productivity scores is that they fail to capture the full picture. A radiologist who reports a high number of cases quickly but makes several errors can still achieve an excellent productivity score, while a colleague with fewer cases but flawless analysis and reports may be considered a poor performer. This emphasis on volume risks encouraging radiologists to prioritize speed over quality, which is a dangerous trend in medicine. We need to move beyond volume metrics and integrate quality, at the image- and diagnosis-level, into how we assess radiologist performance. But how do we do that ? There are options, like peer reviewing cases to assess quality, but this takes time and expertise—two things that are often in short supply in radiology departments. While peer review can work in the short term, sustaining it over time is difficult due to the sheer volume of cases and the limited availability of expert radiologists. This is where I see the future: AI-driven QAPS (i-QAPS). AI can automatically review images and reports simultaneously, flagging discrepancies and alerting radiologists to recheck their cases. These alerts can then feed into a QAPS, providing a more accurate reflection of the radiologist’s true output—one that balances productivity with quality. Some startups are already pioneering this space, such as Avicenna.Ai with their pulmonary embolism detection software, oxipit with ChestEye for chest X-rays, more recently a2z Radiology AI for abdominal and pelvic CT, and earlier Covera Health. These tools are leading the way toward automated quality assessment, enabling us to incorporate the all-important quality factor into productivity scores. So, what about you ? Does your institution monitor radiologist productivity ? Do you integrate image-level quality assessments ? I’d love to hear your thoughts and experiences on this important topic. #Radiology #Quality #Productivity #Performance #Score #AI

29

3 Comments

A good article about the need to build patient trust in Generative AI. One method not mentioned is allowing for the patient to "opt-out" for their own exposure to the tool. Dynamic Global Healthcare/Rev Cycle Staffing What Hospitals Need to Know about Patient Trust in GenAI - MedCity News https://buff.ly/3XhiDDT #healthcare #dallasfortworth

1

A great read about the shift in Epic's strategy, moving from a protectionist model to a true platform. Epic's new approach of building close partnerships a select, highly curated group of tech companies coupled with the enablement of consumer apps is the winning strategy moving forward. This is a 180 degree shift from the platform strategy that Salesforce used to dominate the CRM market - being the platform that every other company must build on. The "Epic model", made possible by AI, instead allows others to build and then feed into the Epic UI. Epic is the solution that the user knows and becomes indispensable in a world where it could have easily been relegated to a database in the background. It further derisks Epic's need to build its own AI capabilities and might even lead the company to a M&A strategy in the future. https://lnkd.in/gtSXNuru

20

Introducing the New Predict User Interface! We are thrilled to unveil the latest enhancement to the Predict Suite—our brand-new AMS User Interface (UI)! This update is designed to elevate your experience when accessing the clinical and financial analytics that have powered our industry for over 20 years. Key Features of the New Predict UI: **Improved Usability & Accessibility: A design that makes navigating easier for all users. **Effortless Navigation: Streamlined paths to the information you need. **'Favorite' & 'Share' Features: Save and share key insights with ease. **Personal Dashboard: Highlight the most important information for your work. **Enhanced Searching Capabilities: Includes auto-completion for faster results. **Consistent Look Across All Modules: A cohesive and professional design throughout. Try it Today: Access the new UI by clicking the Help button in the lower left corner of the current interface, then select Switch to New View. Join Us for an Exclusive Webinar! Discover the latest features live and in action at our upcoming Predict UI webinar. Date: Wednesday, December 4th, 2024 Time: 3:00 PM EST Questions? Contact Michelle Falcon at [email protected] or email [email protected]. AMS (Advanced Medical Strategies)

54

4 Comments

Exciting news from Australian Radiology AI company Annalise.ai : they've secured Medicare’s New Technology Add-on Payment (NTAP) for their obstructive hydrocephalus detection and triage AI software on head CT ! But what does this mean ? 👇 In the US, NTAP provides temporary Medicare payments for new technologies that aren't covered by existing bundled payments for inpatients, and which offer significant clinical benefits. Novelty, high cost, and clinical impact are the three key criteria for qualifying for this payment. This additional reimbursement lasts approximately three years and serves as an incentive for hospitals to adopt and test new technologies that bring clinical value, paving the way for longer-term adoption. According to attorney and radiology policy expert Tom Greeson, NTAP will result in an add-on payment of around $241.39 (65% of the average cost of the technology) per case using Annalise's triage software. This isn't the first time a radiology AI startup has secured such reimbursement. Viz.ai previously received NTAP for its AI product, Viz LVO, which detects large vessel occlusions on CTA during stroke workup. Germany has also introduced a similar framework through the DiGA Fast Track Process for low to medium-risk medical devices, enabling temporary reimbursement until manufacturers provide evidence of clinical benefit. These frameworks are crucial in accelerating AI adoption and generating robust evidence to support long-term financing of AI tools. This achievement marks an important milestone for Annalise.ai and for radiology AI, bringing us closer to broader adoption of this transformative technology. You can read more about this development in Marty Stempniak detailed coverage on Radiology Business in the comment below. Many thanks to Marty for his excellent reporting on major news and trends in radiology, AI, and business. Have a great Friday ! 😃 #Radiology #AI #NTAP #reimbursement

157

8 Comments

Collaborative Care not Hybrid or In-Person Care Virtual First is the framework to transform patient experience and outcomes using data-driven innovation to reimagine health and healthcare operations. This will impact staffing, facilities, technologies, and policy as collaborative care changes the frequency, nature, and value of care interactions. VF goes beyond optimizing encounters or hybrid care; it's about utilizing the most appropriate care mode to address needs at any point in the patient’s health journey. Health is not a single event or a series of disconnected occurrences. Traditionally, healthcare has been plagued by silos within and across systems, specialties, and care modes, resulting in fragmented workflows, communication, and data accessibility. Even within offices, experiences can be episodic, lacking continuity, leading to redundant questions and unnecessary steps. By leveraging data and insights, Virtual First supports seamless care continuity through ongoing collaboration, regardless of care mode. Virtual touchpoints are standard and expected. Good system design incorporates and utilizes virtual tools effectively for the patient, clinical staff, and payers. Virtual First encourages ongoing collaboration to support daily health decisions and needs, shifting focus away from the limited time people spend as "patients." The goal is not merely to offer convenience but to improve health and healthcare delivery by using the most appropriate mode of care, provided by the right person, at the right time, for the right need. Rather than debating the value of virtual vs. hybrid vs. in-person, Virtual First emphasizes collaborative care. Data supports care delivery by any mode at each stage of the health journey. Collaboration can occur between patient-provider, among providers, healthcare team-patient, with caregivers and others. Technology and data allow patient and clinical needs to start to be addressed at the earliest opportunity. Each collaboration informs the next most appropriate step, moving towards a more proactive and person-centric model. By design, Virtual First enables nudges, reminders, and communications as part of a trusted health relationship, not an unfamiliar service. The same virtual tools can be used for chronic disease management, acute care, mental health, preventive initiatives, and risk prevention without data silos or redundancy. Over time, the data collected will support the refinement of evidence-based practices, moving beyond current underperforming care models. Let's not simply reengineer current processes or create new silos. By embracing Virtual First, we can reinvent healthcare to not only be more responsive and efficient, but to address the increasing challenges of access, outcomes, equity, and cost. #VirtualFirst #ReInvent

19

This article offers insights into what radiology innovators need to understand when seeking investment for AI and digital health innovations, detailing the current funding climate and investor criteria. 1️⃣ The diagnostic radiology market is projected to grow to $34.6 billion by 2028, driven largely by AI and digital health technologies. 2️⃣ For AI in radiology, understanding how the innovation integrates into existing clinical workflows is essential, as solutions must add value without disrupting care delivery. 3️⃣ Radiology innovators should target specific customer segments early, such as device manufacturers or large healthcare systems, to maximize initial adoption and revenue generation. 4️⃣ Strategic partnerships, particularly with medical device companies, can be crucial for embedding AI solutions into existing products and reaching a broader market. 5️⃣ The business model should consider whether radiology solutions will be reimbursed by payers or treated as operational costs, influencing pricing and market strategy. 6️⃣ Regulatory and reimbursement strategies are critical for the success of AI-based radiology innovations, as they impact investor willingness and product adoption timelines. 7️⃣ The current venture funding climate is challenging, with a decline in digital health investments, making it vital for innovators to clearly demonstrate efficient business models and strong value propositions. 8️⃣ Investors are increasingly prioritizing businesses that minimize cash burn and can achieve cash flow breakeven quickly, reflecting a shift from the "growth at any cost" mentality seen in 2020-2021. 9️⃣ The importance of investor relationships is emphasized, as long-term partnerships with investors who understand the business can significantly impact future funding rounds and overall success. 🔟 Entrepreneurs should be cautious about seeking high early valuations, as this can lead to challenges in later financing rounds, potentially harming the company’s long-term prospects. ✍🏻 Gaye Bok. Investing in Artificial Intelligence and Digital Health—What Radiology Innovators Need to Know. J Am Coll Radiol. 2024. DOI: 10.1016/j.jacr.2024.06.019

241

13 Comments

I often hear that Big Tech focuses solely on infrastructure and cloud solutions in healthcare, leaving clinical applications (like those used in workflows or patient care) to traditional HealthIT companies and startups. While it's true that Big Tech lacks the detailed workflow and clinical expertise of these players, Microsoft’s recent moves tell a different story. 👇 The latest Forbes article by Dr. Sai Balasubramanian, M.D., J.D. (link in the comment) highlights how Microsoft has launched models capable of generating preliminary reports for chest X-rays, alongside foundational tools that empower healthcare institutions to build their own tailored applications. As a physician, this evolution in HealthTech is particularly exciting to me. I’m often frustrated with the outdated, clunky systems we rely on in healthcare. If we could achieve the same seamless user experience we enjoy with personal tech, it would be transformative for healthcare professionals. I recommend reading this great article about Microsoft’s latest move in healthcare by Dr. Sai Balasubramanian. What do you think—could you see Microsoft becoming your provider for RIS, PACS, and Radiology AI ? #HealthTech #AI #Radiology #BigTech Microsoft - Satya Nadella

22

3 Comments

A solid read on the State of the Union on AI in Healthcare. This observation piqued my attention. "While the hype might seem like a repeat of past waves of short-lived techno-optimism in healthcare, the data tells a different story. " How many of us have gone through the "Trough of Disillusionment" in the hype cycle? This cycle feels different. Check out a recent whitepaper Innovative Renal Care produced on how AI is changing Renal Care - search for "Transforming Renal Care with InnovAItive AI." Is the one time the results might exceed expectations? #aiforgood #aiinhealthcare #aiinrenalcare https://lnkd.in/gPYcZNP3 Bessemer Venture Partners

12

1 Comment

🌟 Thank You to Dr. Joshua Tamayo-Sarver, MD, PhD, FACEP, FAMIA Trustworthy Technology and Innovation in Healthcare Book Series Consortium (TTIC) advisory board member🌟 Happy to acknowledge this insightful comment from our esteemed AB Member, repeat coauthor, and editor of our forthcoming TTIC Venture book, Dr. Josh Tamayo-Sarver. Your words beautifully capture the essence of what we strive to achieve at TTIC. 🚀 Understanding healthcare technology is almost impossible for those outside the field. It requires not just technical expertise, but a deep, integrated understanding of healthcare IT, clinical care, regulation, and the complex business landscape. This is why so many technology companies, regardless of their size, default to failure when attempting to break into #healthcareIT. The challenges are immense, and success demands more than just innovation—it requires comprehensive integration and a relentless commitment to excellence across all these domains. 🏆 This is where TTIC makes a difference. We bring together member experts, industry leaders, providing them with the support, insight, and collaborative environment needed for them to succeed which brings dramatic benefits to their employers. TTIC’s unique approach ensures that innovations are not only cutting-edge but also practical, ethical, and aligned with the realities of healthcare. 🙏 Thank you, Dr. Josh Tamayo-Sarver, for your continued dedication and invaluable contributions. Your leadership and vision are instrumental in driving the future of healthcare technology. Let’s continue to create meaningful advancements that truly transform healthcare. Together, we’re making the impossible possible. Check out the blog on where the field of Healthcare #AI must go: https://lnkd.in/gwCRwwHa #HealthcareIT #AI #Leadership #Innovation #TTIC #Medigram #HealthcareTechnology

8

2 Comments

💡Part 2 of Steve Gravely and Gravely Group's ongoing #HTI2ProposedRule series delving into the Proposed Rule for Health Data, Technology, and Interoperability: Patient Engagement, Information Sharing, and Public Health Interoperability. _______________________________________________________________ See Part 1 here: https://lnkd.in/esuG2yVV vi. New exception- #ProtectingCareAccess. This new exception is specific to EHI related to “reproductive health” as defined under #HIPAA. The exception would protect practices to withhold EHI about reproductive health services based on the Actor’s good faith belief that sharing the EHI could expose the patient to legal risk. This exception is specifically targeting situations in which a woman obtains reproductive health services in a state where such services are legal but their information is requested by someone else. In order to assert this exception, an Actor must demonstrate its good faith belief that the practice is no broader than necessary to protect the patient from legal risk. In addition to this threshold requirement, the Actor must show that the practice satisfies either the patient protection condition or the access to care condition: 1. Patient protection condition requires that EHI being withheld shows that patient obtained reproductive health services or inquired about them or EHI has specific demographic information that implicates reproductive health.   2. Access to care condition are practices that actor takes to limit exposure of providers to legal liability. vii. New exception- #RequestorPreferences. ONC notes that sometimes, a requestor of EHI will state specific preferences that limit the EHI it wants. This could be a date range, a specific type of care or anything else. ONC states that a requestor often does not want all of the EHI that an Actor has in its system. 🎙️Steve Gravely says, "Apparently, ONC has heard concerns that honoring these limitations could be an information blocking violation. I am skeptical, since information blocking is about fulfilling a request and if the requestor limits the scope of the request, I do not see how complying with this could be information blocking. Creating a new exception seems way over the top and makes me wonder if something else is going on." viii. #TEFCAMannerException - ONC is seeking input in HTI-2 about the new TEFCA Manner Exception that it created in HTI-1. Specifically, ONC wants feedback about the “carve-out” related to FHIR APIs. The TEFCA Manner Exception, as adopted in HTI-1, did not apply if the request for EHI was made using the standards adopted in 45 CFR 170.215 which include the FHIR API standards. ONC is seeking comments on whether specifically referring to the standards at 45 CFR 170.215 is too limited. Stay tuned for more insights!

2

So many stumbling stones… Bringing #digitalhealth technologies to patients is not like a sprint - you will stumble repeatedly but need to pick yourself back up to reach success. After all, the gap between regulatory approval and nominal coverage in #healthcare can be up to 5.7 years… The article below is a great summary of findings by Digital Medicine Society (DiMe) on the creativity and tools required to be successful in commercializing DHTs. Several different companies are listed using different regulatory strategies to reach the market. The overarching theme is that an integrated approach between commercial, regulatory, and clinical seems key. “DHTs have unlimited potential to advance health and improve lives, but developers’ ability to bring new products to market and realize their value is contingent on regulatory and downstream decision makers being able to keep up with the pace of innovation and enable patient access to safe and effective technologies.”

31

1 Comment

There’s a lot of buzz around ChatGPT-like interfaces and foundational models in radiology and pathology, but how useful are they in our daily practice ? 👇 I recently watched some impressive demos where a conversational AI analyzes radiology or pathology images, provides diagnoses, and even generates reports. You can ask follow-up questions about the findings, which is undoubtedly impressive. But the big question remains: How does this fit into our workflow? And would we even be willing to pay for it? As radiologists and pathologists, our core task is to review images and deliver a final report. Most of us have a well-optimized workflow, finely tuned with RIS and PACS systems, minimizing unnecessary clicks, loading times, and distractions. Introducing an entirely new tool that requires multiple steps or more time to "discuss" a case could disrupt this efficiency. I can't imagine stepping away from my workflow to spend five minutes chatting with AI for each case—it just wouldn’t be frictionless. That said, there are pain points where conversational AI could be a game changer. Occasionally, we come across a challenging case that sends us to Google, Radiopaedia, or PubMed, trying to find relevant information. This search process is often tedious and inefficient, with hit-or-miss results. Conversational AI could significantly streamline this task. Instead of wading through endless search results, we could describe our findings or even input an image and get targeted, relevant insights much faster. That’s where I see real potential. Another interesting use case is when non-radiologist physicians review images on their own and need quick answers to specific questions. Whether it’s for case-solving or decision-making, having a conversational AI tool that can guide them through the process or offer support without needing a formal radiology consultation could be extremely valuable. This could empower clinicians to make more informed decisions when they don't have direct access to a radiologist. Additionally, This type of AI could be very useful in multimodal analysis for complex patient cases—where various data types come together, and decisions need to be made about the best treatment or test. These tools could be invaluable in multidisciplinary team meetings, helping quickly determine the best next step for patient care. But I’m still left wondering—would you pay for a ChatGPT-like tool for radiology or pathology? How useful would it actually be in your workflow? Video courtesy of Harrison.ai – A warm thank you to their team, as well as to Dimitry Tran and Dr Aengus Tran. #Radiology #Pathology #ConversationalAI #AI #ChatGPT #FoundationalModels #Startups

52

6 Comments

🚧 Triaging and Evaluating AI Initiatives in Healthcare and Life Sciences   Continuing our recent conversations about AI Council (last week - https://lnkd.in/ebammQKA; earlier this week - https://lnkd.in/gBqZrgxw, https://lnkd.in/eU4_Wtv7), let’s dive into a challenge faced by AI Councils: the burden of evaluating and prioritizing AI projects! Amid the GenAI excitement, business leaders often aren't sure if their initiatives need AI/ML or GenAI solutions. This places the responsibility of triaging and clarifying on the AI Council, decelerating adoption.   A structured approach ensures organizations invest in impactful, viable initiatives without overburdening the AI Council. Streamline by implementing GenAI apps at each step. My advanced customers are using tools like 'Amazon Q Apps', a no-code 'Amazon Q Business' feature, to create secure, custom GenAI tools internally. These apps reduce manual work and boost efficiency - https://lnkd.in/eS4Nw2QE Start with self-service. Promote apps like https://lnkd.in/e8hAT8xe for AI Council focus. This lets anyone self-evaluate ideas before presenting. Created on PartyRock, you can recreate it internally using 'Amazon Q Apps'.   Next, create a centralized GenAI app for AI project submissions. It should collect key info like objectives, outcomes, resources, and risks. The app should rapidly evaluate proposals based on: 1/ Organizational alignment, 2/ Patient care/research impact, 3/ Technical feasibility, and 4/ Resource estimates.   For projects passing initial assessment, conduct a thorough analysis covering clinical validation, compliance, ethics, data, integration, and ROI. While leveraging GenAI, human-in-the-loop is crucial. The council should rank projects on strategic importance, impact, readiness, resources, and risk. Engage LoB leaders to validate analysis and set expectations.   Periodically, prioritized projects face AI Council for final approval. The council evaluates: 1/ Clinical Validation: Rigorous real-world testing, 2/ Regulatory Compliance: Navigating FDA, EMA landscapes, 3/ Ethical Assessment: Addressing biases, fairness, and algorithmic transparency.   Next, the council will handle resource allocation. Following an Agile PI planning process, they can map approved initiatives against capacity and budget constraints. This phase should be a dedicated workstream, outsourced to central IT or a partner, to free the council from the details of execution. The workstream owner will manage project milestones, success metrics, reporting, and delivery.   What's your experience with AI Council?   📢 Subscribe to my newsletter: https://lnkd.in/g3bdneR7   #genai #ai #aws #HealthcareAI #AIinLifeSciences #AIStrategy #DigitalHealth

10

Equitable access to healthcare AI is essential to ensure that technological advancements benefit ALL, not just the privileged. Healthcare disparities are already persistent, and AI has the potential to either eliminate or exacerbate those disparities depending on our actions. If access to AI remains uneven, these disparities could worsen, potentially leading to significant differences in patient outcomes and overall public health. Addressing this will require comprehensive, consistent collaboration! #HealthEquity #AI #Healthcare

3

CEP CORE Back Tool: An intuitive decision tree incorporating yellow and red flag screens. Despite being 8 years old, this tool is still relevant due to its structured approach and adherence to best practices. It offers a glimpse into what a clinical decision tree could look like if traditional care delivery is to maintain a viable clinical presence in the future. The tool presents simple, visual data for basic clinical risk analysis and provides foundational guidance for clinical care. By establishing baseline parameters for the clinical community, it paves the way for more optimized processes. Instruments like this, especially in a digitized format, are essential for reducing uncertainty and standardizing the MSK delivery process. This helps MSK professionals establish a core single source of truth, fostering a stronger professional identity. https://cep.health

19

3 Comments

Love it when #digitalhealth gets an evidence boost. The efficacy and safety of new software solutions in #healthcare is often ambiguous due to the convoluted nature of the system. But in certain, less served areas, the data becomes more unambiguous. Take for example #AI-driven clinical decision support systems (CDSS) in maternal care: a recent meta-analysis found that outcomes were improved by 69% with use of these systems. It is important to remark that this is not a single system but different systems in different settings, supporting that the approach works more generally than just being based on a specific health issue or technology (and likely highlighting how low the standard of maternal care is). The systems themselves are relatively simple. Some focus on early detection of potential serious complications, others work on better patient empowerment and education. Just having these tools throughout the pregnancy journey can improve the #patientexperience significantly while achieving better clinical outcomes.

11