Rediminds, Inc

A meaningful milestone for RediMinds, and an even more important signal for where healthcare AI is heading. URAC has named RediMinds among the first organizations to earn its Health Care AI Accreditation, establishing an early benchmark for responsible AI governance in healthcare. For RediMinds, the accreditation recognizes us as both an AI developer and AI user for the technologies supporting our Center for Independent Dispute Resolution. Why does that matter? Because the future of healthcare AI will not be determined by capability alone. It will depend on whether the systems surrounding that capability are built for governance, transparency, risk management, continuous oversight, and accountability. Those disciplines cannot be bolted on after deployment. They have to be part of how AI is designed, implemented, monitored, and improved from the beginning. We’re proud to be among the first organizations independently evaluated against that standard, and even more energized by what it represents: a healthcare AI ecosystem where innovation and trust advance together. Read URAC’s announcement: urac.org/blog/guidehealth-re… #ResponsibleAI #HealthcareAI #AIGovernance #HealthTech #RediMinds #CreateTheFuture

What if an AI system performs worse than a healthcare professional on its own, but still helps that professional make better decisions? It sounds contradictory. But a new study published in npj Digital Medicine reveals why the value of healthcare AI cannot always be measured by comparing the technology against the human. Sometimes, the more important question is what they can accomplish together. Researchers conducted a prospective study in an emergency department, involving 13 radiographers and 2,496 consecutive musculoskeletal examinations over three months. Each examination was independently interpreted by a radiographer, who then reviewed the output of a commercial AI fracture-detection system before recording a final decision. The findings tell an interesting story. ▫️91.6% sensitivity for radiographers working independently, compared with 75.6% for the standalone AI. ▫️97.6% baseline accuracy for radiographers, increasing to 98.0% with AI assistance. ▫️11 errors corrected after AI assistance, while one new error was introduced. ▫️8.8% of examinations received no AI output or an incomplete result. The AI was not more sensitive than the professionals using it. Yet the combined workflow produced a modest, statistically significant improvement in overall accuracy. The improvement came from complementary performance, not algorithmic superiority. That distinction deserves attention. The study also exposed an operational reality: having a capable model is not enough if its output is unavailable, incomplete, or difficult to incorporate into professional judgment. The researchers appropriately caution that the findings come from a single emergency department, the observed behavioral changes were limited, and downstream patient outcomes were not measured. But the underlying principle is highly relevant to the future of healthcare decision infrastructure. AI does not need to become the final decision-maker to create meaningful value. Consider independent medical review. A qualified physician brings clinical knowledge, professional experience, and the responsibility to interpret complex evidence. AI may contribute something different. It can help: → Reconstruct a complicated treatment history. → Surface a finding buried deep in the medical record. → Identify documentation that may contradict an initial interpretation. → Connect clinical evidence to the applicable review criteria. → Highlight gaps requiring additional attention. → Preserve the supporting evidence so the reviewer can independently verify it. These contributions matter even when the technology cannot independently resolve the underlying clinical question. The objective is not to make AI think exactly like the expert. It is to design a workflow in which technology and expertise contribute their respective strengths, while their limitations remain visible. At RediMinds, we build trusted decision infrastructure around this principle. AI helps organize complex evidence and support the review process. Qualified professionals remain responsible for the final determination. And the quality of the system should be evaluated not simply by how often the AI gets something right, but by whether the overall process produces better-supported, more reliable, and more defensible decisions. The new research offers an important perspective on what meaningful augmentation can look like. The future of healthcare AI is not necessarily about building systems that outperform experts. It is about building systems that make expertise more effective. Read the npj Digital Medicine study, published October 27, 2026: nature.com/articles/s41746-0… #HealthcareAI #IndependentMedicalReview #ClinicalAI #DecisionInfrastructure #RediMinds

Most of us were trained to share conclusions, not the thinking behind them. The polished answer. The confident summary. The slide that skips the messy middle. It looks competent, and it hides the part where the real value lives. This week in Thought Fuel Thursday, our Signal Design chapter looks at a forward-looking idea: "The future belongs to the teams that learn how to think out loud together." Thinking out loud isn't talking more. It's making your reasoning visible while it's still forming: "I'm not sure this assumption holds." "Here's what I ruled out and why." "This is half-formed, but I want to say it before I lose it." When that becomes normal, three things shift. Mistakes get caught in minutes instead of weeks. Knowledge stops sitting quietly in individual heads. And ideas get built by the room instead of judged by it. At RediMinds, we think this is a learnable skill; not a personality trait. It grows from small signals: asking "does this land?" before you're certain, narrating your path instead of only your destination, and letting an unfinished thought stay in the air a few seconds longer. The teams that will shape what comes next won't be the ones with the most polished answers. They'll be the ones who got comfortable thinking in the open. #RediMinds #ThoughtFuelThursday #SignalDesign #CollectiveIntelligence #WorkplaceCulture #Collaboration #Innovation #Leadership

What if the most important capability of healthcare AI isn't knowing the answer? What if it's knowing which question to ask next? For years, medical AI has largely been evaluated by giving models a clinical question, supplying the available information, and measuring whether they produce the correct answer. But medicine rarely works that way. A clinician starts with incomplete information. ▫️They ask questions. ▫️They identify possibilities. ▫️They request additional evidence. ▫️They reconsider their assumptions. ▫️And only then do they reach a conclusion. The process of acquiring evidence is itself part of clinical reasoning. A new study published in Science explores what happens when AI is evaluated through that more realistic process. Researchers at Microsoft developed a sequential diagnostic framework in which an AI system coordinates a simulated panel of medical experts. Rather than receiving a complete clinical case upfront, the system progressively gathers information, proposes diagnoses, requests tests, and revises its reasoning. The researchers evaluated the approach using complex cases from The New England Journal of Medicine and emergency department records from MIMIC-IV-ED. The results were notable: → 83.2% diagnostic accuracy on complex NEJM cases. → 80.3% diagnostic accuracy on emergency department cases. → Participating generalist physicians achieved 36.1% and 46.8%, respectively, under the study's comparison conditions. The researchers also evaluated the estimated cost of diagnostic investigations, showing that the system could balance accuracy with more selective information gathering. But the limitations are essential. These were controlled, text-based simulations; not real patient encounters. The findings do not establish superiority over specialist physicians, multidisciplinary clinical teams, or real-world medical practice. And the study examined diagnostic reasoning, not independent medical review. Still, it raises a fascinating question for healthcare decision infrastructure. Should AI simply process the evidence it receives, or should it help determine what evidence the decision actually requires? Consider a complex independent medical review. ▫️The record may contain hundreds of pages. ▫️The applicable guidelines may be clear. ▫️The question under review may be precisely defined. But before a qualified expert can reach a defensible determination, the workflow may need to establish: → Which clinical facts are essential to this specific decision? → What evidence supports the requested treatment or service? → Which findings contradict the proposed conclusion? → Is the applicable clinical criterion actually satisfied? → Is missing documentation preventing a reliable determination? → Would additional information materially change the assessment? → When does the case require specialized clinical interpretation? The distinction is important. An AI system that summarizes everything in the record may reduce reading time. A system that helps identify the next necessary question can make the review process more purposeful. That does not mean AI should independently determine what care is appropriate. It means technology can support the disciplined process through which qualified professionals reach those decisions. At RediMinds, we build trusted decision infrastructure to help healthcare experts navigate complex evidence, apply relevant criteria, and make better-supported determinations. The goal is not to replace professional reasoning with automated conclusions. It is to give experts a clearer view of the evidence, the unanswered questions, and the reasoning their judgment must resolve. This new research points toward a broader direction for healthcare AI. Intelligence is not just the ability to answer a question. It is also the ability to recognize which question must be answered next. And in consequential healthcare decisions, knowing the difference can be just as valuable as the answer itself. Read the Science study, published October 22, 2026: Sequential diagnosis with language models - doi.org/10.1126/science.aeh7… #HealthcareAI #ClinicalReasoning #IndependentMedicalReview #DecisionInfrastructure #ResponsibleAI #RediMinds

When a healthcare system reduces the number of procedures performed, is that evidence of better decision-making? Not necessarily. A reduction in healthcare utilization can tell two very different stories. One is about preventing unnecessary care. The other is about making appropriate care harder to access. A new study published yesterday in JAMA brings this distinction into focus. Researchers from the University of Minnesota and Brown University examined changes in medical service utilization following the January 2026 introduction of Medicare's Wasteful and Inappropriate Service Reduction (WISeR) model. The six-year pilot introduced prior authorization requirements for selected services, encouraging participating organizations to use artificial intelligence in the review process. Using Epic Cosmos electronic health record data, researchers examined the first six months of implementation. The findings were significant: → 15.0% reduction in encounters for epidural steroid injections. → 19.5% reduction in encounters for other targeted services. → No corresponding change in services whose prior authorization requirements had not yet been implemented. The declines were strongest initially and appeared to moderate over time. But here is the question the numbers cannot answer: Was the care that disappeared unnecessary, or was some of it appropriate care that patients could no longer easily receive? The observational analysis does not establish that AI itself caused the reductions, nor does it establish whether the changes improved patient outcomes. And that uncertainty matters. Because prior authorization exists at the intersection of two legitimate healthcare objectives: ▫️Preventing inappropriate utilization. ▫️Protecting timely access to medically necessary care. Optimizing one without measuring the other risks mistaking operational efficiency for better healthcare. For organizations building AI-assisted prior authorization and medical review infrastructure, the implications go beyond processing applications faster. We should be asking: → Are clinical criteria applied accurately and consistently? → Can clinicians understand what evidence supports a determination? → Are requests for additional documentation necessary and actionable? → Are appropriate cases being approved without avoidable delays? → What do appeals and overturned determinations reveal about initial decision quality? → Can the system distinguish genuine improvements in appropriateness from barriers to access? These are the questions that turn utilization management into a measurable decision-quality discipline. At RediMinds, we build AI-enabled infrastructure to support evidence-based healthcare reviews, bringing structure, traceability, and qualified professional judgment into complex decision workflows. Our objective is not simply to help systems process more decisions. It is to help make those decisions more consistent, defensible, and appropriately supported by clinical evidence. The new research is an important reminder that changing healthcare behavior is not the same as improving healthcare outcomes. Both payers and providers have a stake in getting that distinction right. Because: The goal of intelligent utilization management is not less care. It is the right care, supported by the right evidence, at the right time. Read the JAMA research letter: jamanetwork.com/journals/jam… #HealthcareAI #PriorAuthorization #MedicalNecessity #HealthcareQuality #DecisionInfrastructure #RediMinds

Healthcare AI has a reassuring phrase: “There is always a human in the loop.” It sounds like a safeguard. The AI produces an output. A qualified professional reviews it. The professional makes the final decision. But there is a question we don't ask nearly enough: What happens when the AI is wrong, and the human doesn't catch it? A new systematic review and meta-analysis published in npj Digital Medicine examines precisely this problem. Researchers analyzed 46 reports representing 41 independent cohort clusters, studying how clinicians respond to incorrect automated clinical advice. The findings challenge an assumption at the heart of many healthcare AI implementations. → Across three adjusted cohorts, the odds of clinicians reaching the correct conclusion were 10.19 times higher when automated advice was correct than when it was incorrect. → In four real-world studies of computerized ECG interpretation, an estimated 26.6% of erroneous automated interpretations were not recognized or corrected. → Studies involving contemporary AI and large language models also documented harmful reliance on incorrect outputs, although their methods were too heterogeneous to support a pooled estimate. Importantly, the ECG findings varied substantially between studies. They should not be interpreted as a universal error rate for healthcare AI. But the underlying message is difficult to ignore. Human oversight is only effective when it can actually intercept errors. That distinction has enormous implications for independent medical review and other consequential healthcare decisions. Consider an AI-assisted medical necessity determination. The system organizes clinical evidence, identifies applicable criteria, and prepares information for the reviewer. The reviewer remains responsible for the determination. ▫️But what if the AI overlooked a contraindication? ▫️What if it misunderstood the treatment history? ▫️What if it matched the evidence to an inappropriate clinical criterion? ▫️Or presented an incorrect conclusion with convincing supporting language? A human signature at the end of that process doesn't automatically make the decision safe. The infrastructure needs to support meaningful scrutiny. That means asking: → Can the reviewer independently examine the original evidence? → Are important assumptions and uncertainties visible? → Can the professional easily challenge or override an AI-generated interpretation? → Are disagreements captured and examined rather than treated as workflow friction? → Can the organization measure which errors were detected, corrected, or missed? The last question may be the most important. Because a system that tracks how often experts approve AI recommendations is measuring acceptance. A system that measures how reliably experts catch AI mistakes is measuring something much closer to oversight effectiveness. At RediMinds, we build trusted decision infrastructure that combines AI-enabled evidence processing with qualified professional judgment. Our objective is not to create a workflow where humans simply confirm what the technology produces. It is to help experts make decisions they can examine, challenge, support, and defend. The research reinforces an essential principle: Human accountability cannot be reduced to human presence. A professional must remain responsible for the final determination. But the system must also be designed to make that responsibility meaningful. Because: The strongest human-in-the-loop system isn't the one where a human approves every decision. It's the one that helps the human recognize when the AI should not be approved. Read the npj Digital Medicine systematic review and meta-analysis: nature.com/articles/s41746-0… #HealthcareAI #IndependentMedicalReview #HumanInTheLoop #ClinicalDecisionSupport #ResponsibleAI #RediMinds

A healthcare AI system can score well on a benchmark. And still fail to deliver what healthcare professionals actually need. The problem is not always the model. Sometimes, it is what we choose to measure. A new study published in npj Digital Medicine raises an important question about how we evaluate generative AI in clinical workflows. Researchers at UCLA Health examined 66,200 patient portal messages involving 30 primary care physicians. The study compared two conditions: → 21,073 messages where AI-generated drafts were visible to physicians, who could review and edit them. → 45,127 messages where AI drafts were generated silently, while physicians independently composed their own responses. The researchers evaluated the messages using automated metrics, including BLEU, ROUGE, BERTScore, and readability measures. They then examined how those scores related to blinded physician assessments of clinical effectiveness, potential misunderstanding, and harm. The results revealed an important limitation. Only 11% of AI drafts in the visible-draft group were incorporated into final messages with high semantic alignment. And while a semantic similarity metric correlated with physician quality ratings when AI drafts were visible and edited, that relationship did not hold when physicians wrote independently. In other words: A metric that works for evaluating one AI workflow may not be valid for evaluating another. That distinction has significant implications for healthcare decision infrastructure. Consider an AI system supporting independent medical review. A generated clinical summary might receive an excellent similarity score against a reference document. But does that tell us whether it: → Identified the evidence most relevant to the determination? → Distinguished documented facts from clinical interpretations? → Recognized conflicting findings or missing information? → Applied the appropriate clinical criteria? → Preserved the sources supporting its conclusions? → Helped the qualified reviewer reach a more defensible decision? Not necessarily. The study examined patient communications, not independent medical review, and it did not establish the best evaluation framework for every healthcare application. But its central lesson travels well. Evaluation must be designed around the work AI is actually supposed to improve. At RediMinds, we build AI-enabled decision infrastructure for complex healthcare reviews, where quality depends on much more than producing a plausible answer. AI should help organize records, surface relevant evidence, and make expert review more effective. That means performance must ultimately be assessed in the context of the actual workflow, the evidence requirements, and the responsibilities of the professionals making the determination. A high benchmark score can tell us that a model performed well on a particular test. It cannot, by itself, establish that the system is ready to support consequential healthcare decisions. Because: A score tells you whether AI passed the test. Good evaluation tells you whether it was the right test in the first place. Read the npj Digital Medicine study, published October 21, 2026: nature.com/articles/s41746-0… #HealthcareAI #IndependentMedicalReview #ClinicalAI #DecisionInfrastructure #ResponsibleAI #RediMinds

Some teams move fast because they rush. Others move fast because they trust each other. This week in Thought Fuel Thursday, our Signal Design chapter looks at what makes complex work feel coordinated instead of crowded: "Trust is the shortcut that turns complex work into graceful motion." Trust doesn't replace rigor. It removes the friction around it: the extra approval loops, the repeated status checks, the concerns that wait too long to be said out loud. It's built in small, practical ways: → Commitments kept, or renegotiated early. → Reasoning shared, not just results. → Questions treated as contribution, not suspicion. At RediMinds, this week's invitation is simple: find one place where your team is double-checking out of habit rather than need, and design a clearer way to rely on each other. Speed is useful. Synchronization is better. #RediMinds #ThoughtFuelThursday #SignalDesign #Trust #Collaboration #WorkplaceCulture #Leadership

A clinical standard is supposed to create consistency. But what happens when different systems apply the same standard, and reach dramatically different results? A new study published yesterday in the Journal of the American Medical Informatics Association reveals a problem that extends far beyond sepsis research. Researchers systematically reviewed 64 studies that used the established Sepsis-3 clinical definition to identify cases in two major intensive care databases. The findings were remarkable. Across studies using the MIMIC-III database, reported sepsis detection rates ranged from 3.4% to 65.2%. Even among studies using identical cohort definitions, substantial differences remained: → 16.9% to 42.2% in MIMIC-III. → 13.9% to 31.4% in eICU-CRD. The researchers also examined available source code and identified 321 implementation decisions affecting how the clinical definition was translated into computational logic. These included: → How missing clinical information was handled. → Which time windows were used. → How measurements were aggregated. → How organ dysfunction was calculated. → How suspected infections were identified. And only 17.2% of the studies documented their missing-data handling methods. The study examined retrospective research methods, not the accuracy of individual clinical diagnoses or insurance determinations. But it exposes an important distinction. A standardized definition does not automatically produce a standardized decision. There is an entire layer between the rule and the result. For independent medical review, prior authorization, disability determinations, and other evidence-intensive healthcare processes, that layer deserves serious attention. Consider two reviewers evaluating the same medical necessity question. They may have access to the same clinical record and the same guideline. But consistency also depends on: → Which guideline version applies. → How its criteria are interpreted. → Which documentation satisfies each requirement. → How missing or conflicting evidence is treated. → When an exception requires expert interpretation. → Whether the reasoning can be reconstructed and independently examined. These are not just software implementation details. They are part of the decision itself. At RediMinds, we build trusted decision infrastructure that brings structure, traceability, and qualified human expertise to complex healthcare reviews. AI can help organize evidence and connect clinical documentation with applicable criteria. But a defensible determination also requires clarity about how those criteria are applied, and where professional judgment must take over. The lesson from this research is bigger than any individual algorithm. Writing the standard is only the beginning. The real test is whether the systems applying it can produce decisions that are consistent, explainable, and worthy of trust. Because in healthcare: Consistency isn't achieved by having the same rules. It's achieved by applying them with the same rigor. Read the JAMIA study, published October 20, 2026: academic.oup.com/jamia/advan… #HealthcareAI #IndependentMedicalReview #ClinicalDecisionSupport #HealthcareQuality #DecisionInfrastructure #RediMinds

The day an AI system goes live is not necessarily the day its value becomes clear. Healthcare organizations often evaluate AI through immediate performance: ▫️How much faster is the task? ▫️How much time does the technology save? ▫️How much work can it automate? But there is another question worth asking: What happens after the technology becomes part of the actual workflow? A new JAMIA study offers compelling evidence. Researchers at Washington University School of Medicine and collaborating institutions examined the longitudinal impact of ambient AI scribes on 220 primary care clinicians across 36 clinics and 314,845 patient encounters. Using a quasi-experimental design, they compared documentation and productivity measures over 150 days before and after clinicians received access to the technology. The findings reveal an interesting progression. → Note-writing time decreased by 7% initially, reaching 16% by day 150. → After-hours documentation time decreased by 19% by day 150. → Time spent in the EHR outside scheduled hours decreased by 13%. → Work relative value units increased by 2% at day 150. Importantly, the researchers found no immediate change in after-hours work or productivity when the technology was first introduced. Those differences emerged over time. The study was observational rather than randomized, so the findings should not be interpreted as definitive proof of causation. And it examined AI-assisted documentation, not independent medical review. But the broader lesson is significant. AI performance and operational value are two different measurements. A tool can demonstrate impressive capabilities while delivering limited practical improvement if the workflow around it remains unchanged. For healthcare decision infrastructure, that distinction matters enormously. Consider an independent medical review platform. Its value should not be measured solely by how quickly AI summarizes clinical documents. We should also ask: → Are qualified reviewers spending less time navigating records and more time exercising clinical judgment? → Are cases arriving with fewer documentation gaps? → Are unnecessary handoffs and repeated administrative tasks being reduced? → Are quality and consistency improving alongside throughput? → Does the system continue delivering measurable value after the initial implementation? At RediMinds, we build AI-enabled infrastructure to make complex healthcare decision workflows more efficient, structured, and defensible. The objective is not simply to introduce more automation. It is to create lasting improvements in how evidence is prepared, expertise is applied, and decisions are supported. Because the return on AI is not just what the technology produces. It is what the workflow no longer demands of the people responsible for the decision. Read the study: doi.org/10.1093/jamia/ocag15… #HealthcareAI #HealthcareOperations #ClinicalWorkflow #DecisionInfrastructure #RediMinds

A healthcare network can look complete on paper. Hundreds of clinicians. Multiple specialties. Broad geographic coverage. But one question matters more than the number of names: Can the right expert actually be reached when they're needed? A new study published in JAMA Health Forum reveals just how significant the gap between listed expertise and accessible expertise can be. Researchers conducted 5,822 secret-shopper calls to psychiatrists and psychologists listed in commercial insurance directories across five U.S. states. The findings were striking: ▫️ 61.3% of calls produced no definitive outcome, frequently because callers reached voicemail or automated systems. ▫️ Among calls that could be evaluated, 55.6% encountered directory inaccuracies, including incorrect contact or specialty information. ▫️ Only 23.6% of evaluable calls resulted in an appointment with the listed clinician. ▫️ Across all initial call attempts, just 12.4% secured an appointment with any clinician. The study examined access to mental healthcare, not independent medical review. But it exposes a broader healthcare infrastructure problem. Knowing that an expert exists is not the same as knowing that the expert is available, appropriately qualified, and suitable for a specific need. Consider what that means for independent medical review. A case may require a specialist with precise clinical expertise, current licensure, appropriate board certification, relevant experience, and independence from the parties involved. Identifying that professional is only the beginning. The system must also establish: → Is this the appropriate specialty for the clinical question? → Are the clinician's credentials current and verified? → Are there conflicts that could compromise independence? → Is the expert available within the required timeframe? → Can the case be assigned, reviewed, and completed without unnecessary delays? These are not merely administrative considerations. They are part of what makes a healthcare determination credible. At RediMinds, our approach to trusted decision infrastructure combines AI-enabled workflows with access to credentialed physicians and allied health professionals across more than 50 specialties. Technology can help organize evidence, coordinate reviews, and support expert matching. But the quality of a determination ultimately depends on connecting the right evidence with the right qualified professional. A network should not be measured only by how many experts it lists. It should be measured by how reliably it connects each decision to the expertise that decision requires. Because in healthcare, access to expertise is not a directory feature. It is an operational capability. Read the JAMA Health Forum study: jamanetwork.com/journals/jam… #HealthcareAI #IndependentMedicalReview #ClinicianCredentialing #HealthcareOperations #RediMinds

Healthcare AI teams often talk about trust as though it can be designed into an interface. Show a confidence score. Add an explanation. Attach citations. And the clinician will trust the system appropriately. A new JAMIA study suggests it is more complicated than that. Researchers ran a randomized experiment with 220 clinicians, comparing AI diagnostic support presented in three ways: ▫️ no additional transparency ▫️ numeric confidence scores ▫️ narrative explanations with citations The surprising result? Neither confidence scores nor narrative explanations produced a statistically detectable change in clinicians’ trust in the AI tool, overall trust in AI, or perceived credibility of the information. But something else did change. After receiving AI assistance, clinicians’ confidence in their own answers increased by 6.7% overall. Narrative explanations produced a greater increase in confidence than either no-transparency AI or confidence scores alone. And across the experiment, clinicians changed or refined their diagnoses after seeing the AI output 45.8% of the time. That distinction is important. Trust. Confidence. Behavior. They are not the same thing. An AI system can influence a healthcare professional without making them consciously trust it more. An explanation can make someone feel more confident without necessarily making the underlying recommendation more reliable. For consequential healthcare review, that means the objective should not be to make AI appear more trustworthy. It should be to help experts calibrate their reliance on it. → Can the reviewer see the source evidence? → Can they challenge the system’s interpretation? → Can they distinguish a strong signal from an uncertain one? → Can they recognize when the case needs escalation rather than another AI-generated answer? At RediMinds, human-in-the-loop should mean more than placing a professional at the end of an automated workflow. The system itself should help that professional know when to rely, when to verify, and when to disagree. Because the safest AI may not be the one clinicians trust the most. It may be the one that helps them trust it the right amount. Read the study: academic.oup.com/jamia/advan… #HealthcareAI #MedicalReview #HumanInTheLoop #ClinicalAI #ResponsibleAI #RediMinds

A project plan can tell everyone what to do. Can everyone explain what the work is for? This week in Thought Fuel Thursday, our Signal Design chapter explores the connection between shared purpose and meaningful effort: “A shared purpose gives the brain its cleanest energy source.” Not energy in the biological sense, and not a reason to expect endless enthusiasm. Purpose does something more practical: it gives effort a direction. When people understand the outcome they are working toward, they have a reference point for decisions: What deserves our attention? Which trade-off serves the people using this? What does this contribution make possible next? At RediMinds, this week’s invitation is to make the purpose of a piece of work specific enough to guide it. Try one sentence at your next team check-in: “This work matters because it helps ______ do ______.” Then compare answers. The conversation may be as useful as the sentence. #RediMinds #ThoughtFuelThursday #SignalDesign #WorkplaceCulture

Every complex healthcare decision leaves behind two records. The first is obvious: the clinical record being reviewed. The second is easier to overlook: the path the expert took through it. ▫️Which information did they open first? ▫️What did they return to? ▫️Where did they hesitate? ▫️What caused them to change direction? ▫️Which parts of the workflow created friction? A new JAMIA study offers an interesting glimpse into why those interaction patterns matter. Researchers analyzed more than 295 million EHR audit-log actions from inpatient settings between 2019 and 2024 and trained a language model to learn the sequences behind clinician interactions with the electronic health record. Rather than reducing workflow to measures such as total time spent in the EHR, the researchers modeled the sequence and context of actions. The strongest approach achieved 74.7% next-action prediction accuracy in the initial test period, compared with 42.3% for a simpler model based only on the immediately preceding action. It maintained a substantial advantage across later out-of-sample periods, although performance declined over time. The authors point toward applications in task identification, workflow characterization, and safety-focused analysis. This study was not about independent medical review. But the design principle is highly relevant. A review platform can record the final determination. A decision infrastructure can also help us understand how experts reached it. Applied thoughtfully, and with appropriate governance, workflow signals could help reveal: → where reviewers repeatedly search for the same evidence, → where incomplete records cause backtracking, → which cases create unusual cognitive friction, → when additional expertise or escalation may be needed, → and where the system itself can make the review clearer. That does not mean automating the reviewer. It means learning how to better support them. At RediMinds, we think the future of healthcare review is not just about making individual decisions faster. It is about building systems that can become more useful, more observable, and more defensible with every decision they support. Because: The determination is the output. The workflow tells you how the system is really working. Read the Journal of the American Medical Informatics Association study: academic.oup.com/jamia/advan… #HealthcareAI #MedicalReview #ClinicalWorkflow #DecisionInfrastructure #HumanInTheLoop #RediMinds

Healthcare AI has spent a lot of time asking: How capable is the model? A better deployment question may be: How much model does the task actually need? A new JAMIA systematic review examined small language models, defined as models with 4 billion parameters or fewer, across clinical applications. The findings were deliberately mixed. Some smaller, domain-adapted models performed competitively on particular tasks, and the researchers estimated that a 4-billion-parameter model could be memory-feasible on a single GPU. But the evidence base was still small and heterogeneous, making broad comparisons with larger models premature. The safety evaluation was even more revealing: ▫️hallucination was assessed in only 6 of 11 studies ▫️quantitative calibration or epistemic uncertainty was essentially not evaluated ▫️real on-hardware inference timing was reported in only 2 of 11 studies That distinction matters. In healthcare, the objective should not be to put the largest available model behind every workflow. → Different tasks demand different levels of reasoning, latency, privacy, infrastructure, and human oversight. → Extracting a fact from a record is not the same problem as interpreting conflicting clinical evidence. → Preparing a case is not the same as making the determination. The strongest decision infrastructure should therefore be designed around the task, not the prestige of the model. At RediMinds, that principle matters because AI is there to support qualified experts, not to make complexity disappear behind a bigger black box. The right model is not necessarily the biggest one. It is the one whose performance, limitations, and role in the decision can be understood. Read the systematic review: academic.oup.com/jamia/advan… #HealthcareAI #ClinicalAI #DecisionInfrastructure #ResponsibleAI #RediMinds

Healthcare AI can process enormous amounts of clinical information. But there is a quieter problem underneath the model: What if the clinical language itself is incomplete, inconsistent, or imprecise? A new JAMIA study proposes a quality-assurance framework for SNOMED CT, the clinical terminology used to give structured meaning to diagnoses, findings, procedures, and other healthcare concepts. The researchers organize terminology quality around eight dimensions: ▫️ clarity ▫️ conciseness ▫️ consistency ▫️ correctness ▫️ domain coverage ▫️ currency ▫️ completeness ▫️ conformance They also distinguish between errors caused by something being wrong and errors caused by something being missing. That distinction matters well beyond terminology management. Healthcare decision systems constantly translate between: clinical notes → structured concepts → criteria → guidelines → evidence → decisions. Every transformation creates an opportunity to lose nuance. → A diagnosis can become broader than the clinician intended. → A relationship between concepts can disappear. → An outdated term can survive inside a workflow. → A missing mapping can quietly make relevant evidence harder to find. → And once that structured information feeds an AI system, the model may process the representation perfectly, while the representation itself is imperfect. For independent medical review, prior authorization, disability determinations, workers’ compensation, and other evidence-intensive workflows, AI should therefore do more than normalize clinical language. It should preserve the path back to what the record actually said. At RediMinds, we believe structured data should make expert review clearer without flattening the clinical meaning underneath it. Because a decision can only be as defensible as the evidence, and the meaning, we preserve along the way. Read the Journal of the American Medical Informatics Association study: academic.oup.com/jamia/advan… #HealthcareAI #MedicalReview #ClinicalEvidence #HealthInformatics #DecisionSupport #RediMinds

In healthcare, evidence has a clock. A lab value can be accurate, and no longer relevant. A medication can appear in the record, but already have been stopped. A clinical finding can be real, but precede the decision under review by months. That distinction is easy for a human expert to understand. It is harder to build into AI. A new npj Digital Medicine study published yesterday evaluated an “audit-ready” EHR foundation model across seven health systems and 847,000 high-risk medication orders. What makes the work interesting is not simply the model. It is how the evidence was represented. The system explicitly encoded: ▫️ time ▫️ units ▫️ provenance ▫️ whether information was actually observable ▫️ and structured missingness. In the retrospective evaluation, the model with hard constraints produced 3.2 missed actionable alerts per 1,000 eligible orders, compared with 9.4 for rules alone and 4.8 for the strongest RAG + LLM + rules comparator tested. The authors are appropriately careful: this was primarily an offline process evaluation. It does not establish improved patient outcomes or reduced harm. But the design principle matters far beyond medication safety. For complex medical review, it is not enough for AI to know that a fact exists somewhere in the record. It needs to understand: → when it was true, → where it came from, → what units and context give it meaning, → whether newer evidence supersedes it, → and when the information needed to answer the question simply is not there. A reviewer deciding medical necessity, disability, workers’ compensation, or another consequential healthcare question should not receive a timeless collection of extracted facts. They need an evidence trail. At RediMinds, that is how we think about trusted decision infrastructure: AI should help organize the record without stripping away the context that makes the evidence interpretable. Because: Evidence without provenance can mislead. Evidence without time can mislead too. A defensible decision needs both. Read the npj Digital Medicine study: nature.com/articles/s41746-0… #HealthcareAI #MedicalReview #ClinicalEvidence #DecisionSupport #ResponsibleAI #RediMinds

A good question gives someone else’s expertise somewhere to land. This week in Thought Fuel Thursday, our Signal Design chapter explores how the questions we ask can expand what a team discovers: “Generous questions create smarter answers than confident statements ever could.” A generous question is not less rigorous. It makes room for evidence, experience, and possibilities beyond the asker’s own view. “What are you seeing that I might be missing?” “What would make this worth testing?” “What would change our minds?” These questions invite people to contribute, not simply agree. They give expertise a chance to connect rather than compete. At RediMinds, this week’s invitation is practical: bring your knowledge to the conversation, then ask something that lets another person extend it. The most useful question may be the one you cannot answer on your own. The next time you're in a meeting, a one-on-one, or even a difficult conversation, try replacing your instinct to assert with the discipline to inquire. Ask something you don't already know the answer to. Watch what happens when you give someone's mind room to stretch. #RediMinds #ThoughtFuelThursday #SignalDesign #Collaboration #Curiosity #WorkplaceCulture

We're excited to unveil our new visual identity. Meet the new RediMinds♥️

Healthcare AI is increasingly expected to do more than produce an answer. It is expected to explain how it got there. That sounds like progress. But a new npj Digital Medicine review suggests the explanation layer itself can be surprisingly opaque. Researchers examined 100 peer-reviewed studies using Shapley value explanations, a widely used method for showing which features contributed to an AI prediction. They found: ▫️ 91% did not report the algorithm used to calculate the Shapley explanations. ▫️ 97% did not report the source or size of the reference population used in the calculation. That matters because an explanation can look precise without necessarily being easy to reproduce, interpret, or challenge. And in consequential healthcare decisions, “the model says these factors mattered” is not enough. A trustworthy decision system should make it possible to understand: → what evidence influenced the output, → what assumptions shaped the explanation, → what comparison or baseline was used, → where uncertainty remains, → and whether a qualified expert can interrogate the result. This is particularly important in medical review. A reviewer should not receive a polished explanation and be expected to trust it. They should be able to move from: conclusion → reasoning → evidence → source. At RediMinds, traceability is not something we add after a decision is made. It is part of how the decision should be built. Because: Explainable AI is not just AI that produces an explanation. It is AI whose explanation can itself be examined. Read the npj Digital Medicine study: nature.com/articles/s41746-0… #HealthcareAI #MedicalReview #ExplainableAI #DecisionSupport #RediMinds