HealthRadar editorial summaries

The science behind HealthRadar

Recent evidence on laboratory-result communication, patient understanding, interpretive workflows, and rules-based clinical logic

Last literature review: September 11, 2026

The summaries below were written by HealthRadar to describe why selected publications are relevant to the questions guiding our development. They are not the articles’ official abstracts. Readers should consult the original publications for complete methods, results, limitations, and author conclusions. Inclusion does not imply endorsement of HealthRadar by the authors, institutions, journals, or publishers.

HealthRadar is being developed at the intersection of laboratory medicine, patient communication, clinical reasoning, health informatics, and rules-based medical software. This page summarizes independent research that helps frame those areas. Start with section A for patient understanding and action, section B for clinical and AI workflows, and section C for system safety and repeatability.

Recent research shows that immediate access to laboratory results does not consistently produce understanding; that visual and textual presentation can influence perceived urgency; that generic educational material may not resolve patient uncertainty; and that both generative and rules-based systems require careful evaluation and oversight.

The publications collected here help define the scientific questions HealthRadar is intended to address. They do not constitute clinical validation of HealthRadar.

A. Patient understanding, presentation, and action

Patient understanding, presentation, and actionSystematic review · 2026

Patient Experiences With Online Laboratory Test Presentations From Access to Activation: Systematic Review

Authors: Lustria MLA, Atrinawati L, Aliche O, Kim K, He Z.

Journal/source: J Med Internet Res.

Full citation: Lustria MLA, Atrinawati L, Aliche O, Kim K, He Z. Patient Experiences With Online Laboratory Test Presentations From Access to Activation: Systematic Review. J Med Internet Res. 2026;28:e88259. doi:10.2196/88259.

DOI: 10.2196/88259

HealthRadar editorial summary

This systematic review synthesized 39 studies examining how patients access, interpret, and act on laboratory results delivered through online portals. Immediate-release policies increased result viewing, secure messaging, and follow-up activity, but access did not reliably produce understanding. Patients frequently had difficulty judging borderline findings, distinguishing minor abnormalities from important ones, and determining when follow-up was warranted. Visual and textual cues could improve clarity, but poorly calibrated cues could overemphasize small deviations or fail to communicate meaningful risk. Patient responses were also shaped by health literacy, digital access, language, clinician communication, and the availability of interpretive support. The review strongly supports treating laboratory communication as a continuum from access to interpretation to action, rather than assuming that displaying results completes the communication task.

Why it matters to HealthRadar

This is a current literature-level statement of the problem HealthRadar is attempting to address: laboratory access without reliable contextual interpretation or navigable next steps.

Read the original article

Patient understanding, presentation, and actionCross-sectional survey analysis · 2026

Patient-Reported Experiences With Viewing and Understanding Test Results in Patient Portals: Cross-Sectional Survey Analysis

Authors: Richwine C, Steitz B, Everson J.

Journal/source: J Med Internet Res.

Full citation: Richwine C, Steitz B, Everson J. Patient-Reported Experiences With Viewing and Understanding Test Results in Patient Portals: Cross-Sectional Survey Analysis. J Med Internet Res. 2026;28:e94098. doi:10.2196/94098.

DOI: 10.2196/94098

HealthRadar editorial summary

Using nationally representative 2024 Health Information National Trends Survey data, this study examined 6,045 US adults who had recently received test results. Seventy percent viewed results in a patient portal, and 58% viewed them before hearing from a healthcare professional. Only 28% reported being offered a choice about whether to receive results before clinician communication, and just 66% said they understood both the results and their implications for care. Viewing and understanding varied with education, digital literacy, and clinician encouragement to use the portal. The study demonstrates that immediate electronic release has become routine while meaningful comprehension remains uneven. It also suggests that interpretation cannot be separated from communication preferences, anticipatory guidance, and disparities in patients’ ability to navigate digital health information.

Why it matters to HealthRadar

The study provides contemporary national evidence for the gap between result availability and usable clinical understanding.

Read the original article

Patient understanding, presentation, and actionLongitudinal qualitative study · 2026

Making non-emergent abnormal results actionable: a short-term longitudinal qualitative study in two hospital-affiliated health examination centers in China

Authors: Yao L, Deng Z, Luo F, He L.

Journal/source: Front Public Health.

Full citation: Yao L, Deng Z, Luo F, He L. Making non-emergent abnormal results actionable: a short-term longitudinal qualitative study in two hospital-affiliated health examination centers in China. Front Public Health. 2026;14:1900806. doi:10.3389/fpubh.2026.1900806.

DOI: 10.3389/fpubh.2026.1900806

HealthRadar editorial summary

This longitudinal qualitative study followed 48 adults who had received non-emergent abnormal laboratory results; 28 completed follow-up interviews 28–43 days later. Participants typically noticed that a result was abnormal before they understood its meaning. Interpretation then unfolded through personal history, family discussions, internet searches, risk perceptions, practical barriers, and attempts to identify an appropriate source of care. Five stages emerged: recognizing the signal, constructing meaning, appraising risk, navigating follow-up, and responding. Communication was considered more actionable when it was respectful, proportionate to the apparent concern, explicit about uncertainty, time-bound, and linked to a realistic service route. The authors concluded that patients were not simply missing information; the surrounding system had transferred substantial interpretive and navigational work to patients and families.

Why it matters to HealthRadar

The findings support HealthRadar’s emphasis on priority, proportional urgency, understandable meaning, and one navigable next step.

Read the original article

Patient understanding, presentation, and actionSystematic review · 2024

Enhancing Patient Understanding of Laboratory Test Results: Systematic Review of Presentation Formats and Their Impact on Perception, Decision, Action, and Memory

Authors: van der Mee FAM, Schaper F, Jansen J, Bons JAP, Meex SJR, Cals JWL.

Journal/source: J Med Internet Res.

Full citation: van der Mee FAM, Schaper F, Jansen J, Bons JAP, Meex SJR, Cals JWL. Enhancing Patient Understanding of Laboratory Test Results: Systematic Review of Presentation Formats and Their Impact on Perception, Decision, Action, and Memory. J Med Internet Res. 2024;26:e53993. doi:10.2196/53993.

DOI: 10.2196/53993

HealthRadar editorial summary

This systematic review evaluated 18 studies of formats used to present numeric laboratory results to patients. Most studies relied on simulated rather than real-world results, and the evidence base was heterogeneous. Horizontal bars showing a result in relation to a reference or individualized goal range generally improved perceived comprehensibility, usability, and satisfaction compared with numbers and reference ranges alone. Some visual formats also reduced participants’ stated intention to search elsewhere or contact a physician. However, behavioral action and memory were rarely evaluated, making it impossible to identify one universally preferred format or determine whether immediate comprehension produced sustained appropriate behavior. The review demonstrates that presentation format is not decorative: it changes how patients perceive magnitude, meaning, and potential concern. It also underscores the need to test displays in real clinical settings and with varied patient populations.

Why it matters to HealthRadar

A clinically sound interpretation can still fail if its presentation produces confusion, alarm, or false reassurance.

Read the original article

Patient understanding, presentation, and actionQualitative think-aloud study · 2025

False Calm and False Alarm: A Qualitative Study of Confusion and Misinterpretation of a Laboratory Results Graph

Authors: Monkman H, Schmit A, Nyholt D, MacDonald L, Lesselroth B.

Journal/source: Stud Health Technol Inform.

Full citation: Monkman H, Schmit A, Nyholt D, MacDonald L, Lesselroth B. False Calm and False Alarm: A Qualitative Study of Confusion and Misinterpretation of a Laboratory Results Graph. Stud Health Technol Inform. 2025;326:106-110. doi:10.3233/SHTI250248.

DOI: 10.3233/SHTI250248

HealthRadar editorial summary

In this think-aloud study, 24 adults interpreted a longitudinal graph of laboratory values modeled on a community laboratory portal. Eighteen participants expressed confusion, and 12 misinterpreted at least one feature. Problems included unclear axis labeling, incorrect interpretation of dates or the number of observations, and uncertainty about whether a change was clinically meaningful. Some users inferred concern where the graph did not support it, while others discounted potentially relevant changes. The findings show that a trend display can appear intuitive while still producing materially different interpretations among viewers. Longitudinal laboratory communication therefore requires more than plotting values: dates, scales, reference context, clinical significance, and uncertainty must be made explicit, and actual users should test the display before deployment.

Why it matters to HealthRadar

This study has direct implications for TrendLens. A visually attractive graph is not automatically an understandable or safe graph.

Read the original article

Patient understanding, presentation, and actionQuality-improvement interrupted time-series study · 2025

Patient-Friendly Test Results and Patient-Initiated Messaging Among Adult Outpatients

Authors: Steitz BD, Guide A, Rodriguez K, et al.

Journal/source: JAMA Netw Open.

Full citation: Steitz BD, Guide A, Rodriguez K, et al. Patient-Friendly Test Results and Patient-Initiated Messaging Among Adult Outpatients. JAMA Netw Open. 2025;8(11):e2543879. doi:10.1001/jamanetworkopen.2025.43879.

DOI: 10.1001/jamanetworkopen.2025.43879

HealthRadar editorial summary

This quality-improvement study evaluated the addition of patient-friendly descriptions, basic interpretation, graphical reference information, and follow-up reassurance to commonly viewed laboratory results in an electronic portal. The analysis included 205,139 patients who reviewed 829,902 results during 2024. The proportion of results followed by a patient-initiated message within 24 hours changed only from 17.5% to 17.3%, which was not clinically meaningful overall. A modest 0.8-percentage-point decrease occurred for tests ordered in primary care, while some other categories showed no benefit or increased messaging. The intervention was informative but largely generic. The findings suggest that standard educational text can support transparency without necessarily resolving the patient’s case-specific questions, uncertainty, or desire for clinical context. More individualized and workflow-aware approaches may be needed.

Why it matters to HealthRadar

The findings suggest that definitions and boilerplate education alone are unlikely to fill the interpretation gap. Contextual synthesis may be the missing layer.

Read the original article

B. Interpretation within clinical and AI workflows

Interpretation within clinical and AI workflowsProspective quality-improvement study · 2025

Clinician Perspectives on AI-Generated Drafts of Patient Test Result Explanations

Authors: Shah SJ, Nair A, Murtagh K, et al.

Journal/source: JAMA Netw Open.

Full citation: Shah SJ, Nair A, Murtagh K, et al. Clinician Perspectives on AI-Generated Drafts of Patient Test Result Explanations. JAMA Netw Open. 2025;8(8):e2528794. doi:10.1001/jamanetworkopen.2025.28794.

DOI: 10.1001/jamanetworkopen.2025.28794

HealthRadar editorial summary

Stanford Health Care integrated a generative-AI tool into the electronic health record to draft clinician-reviewed explanations for laboratory, imaging, and pathology results. Of 244 clinicians who used the system, 93 completed follow-up surveys. Approximately 85% found it easy to use, 72% considered it useful for laboratory results, 71% reported improved efficiency, and 72% perceived higher-quality explanations. Nevertheless, clinicians identified important shortcomings involving factual accuracy, completeness, omitted abnormal results, insufficient clinical context, result sequencing, workflow integration, and the need for editing. Only slightly more than half considered the system ready for broad implementation. The study shows that interpretation can be embedded at the point of result release and may improve clinician workflow, but favorable usability ratings do not eliminate the need for clinician review, context integration, quality control, and explicit governance.

Why it matters to HealthRadar

This is a real-world demonstration of the workflow HealthRadar may eventually enter—and of the limitations of unconstrained generation within that workflow.

Read the original article

Interpretation within clinical and AI workflowsComparative evaluation study · 2024

Comparison of ChatGPT, Gemini, and Le Chat with physician interpretations of medical laboratory questions from an online health forum

Authors: Meyer A, Soleman A, Riese J, Streichert T.

Journal/source: Clin Chem Lab Med.

Full citation: Meyer A, Soleman A, Riese J, Streichert T. Comparison of ChatGPT, Gemini, and Le Chat with physician interpretations of medical laboratory questions from an online health forum. Clin Chem Lab Med. 2024;62(12):2425-2434. doi:10.1515/cclm-2024-0246.

DOI: 10.1515/cclm-2024-0246

HealthRadar editorial summary

The investigators submitted 100 authentic patient questions about complete blood count results to ChatGPT, Gemini, and Le Chat and compared the responses with answers provided by certified physicians on an online health forum. Although the chatbots often communicated with greater apparent empathy, their interpretations were inferior to physician responses and frequently became overly generalized or erroneous when questions required contextual clinical reasoning. Only 51%–64% of chatbot responses were judged appropriate, while 22%–33% overestimated the likely seriousness of the patient’s condition. The models regularly included disclaimers and recommendations to consult a professional, but those safeguards did not correct the substantive weaknesses. The study illustrates the difference between a response that sounds trustworthy and one that is clinically proportionate, accurate, and context-sensitive.

Why it matters to HealthRadar

The study directly supports evaluating medical interpretation on clinical substance, proportionality, data fidelity, and urgency—not elegance or empathy alone.

Read the original article

C. Governance, malfunction risk, and consistency

Governance, malfunction risk, and consistencyScoping review · 2024

A scoping review of rule-based clinical decision support malfunctions

Authors: Thayer JG, Franklin A, Miller JM, Grundmeier RW, Rogith D, Wright A.

Journal/source: J Am Med Inform Assoc.

Full citation: Thayer JG, Franklin A, Miller JM, Grundmeier RW, Rogith D, Wright A. A scoping review of rule-based clinical decision support malfunctions. J Am Med Inform Assoc. 2024;31(10):2405-2413. doi:10.1093/jamia/ocae187.

DOI: 10.1093/jamia/ocae187

HealthRadar editorial summary

This scoping review identified 28 publications describing 130 malfunctions involving rule-based clinical decision-support systems. Failures occurred across stand-alone calculators, integrated alerts, and service-oriented architectures. Causes extended beyond incorrect clinical logic to include implementation defects, data-source problems, hardware errors, organizational-policy changes, interoperability failures, and sociotechnical factors. Only 29 malfunctions—22%—described the potential effect on patient care, suggesting that malfunction reporting often focuses on technical behavior without adequately characterizing clinical consequences. The review is an important counterweight to the idea that deterministic systems are inherently safe because their outputs are reproducible. Rules provide inspectability and testability, but they still require controlled versions, input validation, regression cases, data-lineage checks, incident monitoring, and explicit assessment of downstream clinical impact.

Why it matters to HealthRadar

Deterministic behavior is an advantage only when the entire rule-to-data-to-output pathway is governed and monitored.

Read the original article

Governance, malfunction risk, and consistencyPREPRINT — NOT PEER REVIEWEDPreprint — not peer reviewed

A statistical framework for evaluating the repeatability and reproducibility of large language models

Authors: Shyr C, Ren B, Hsu C-Y, et al.

Journal/source: medRxiv.

Full citation: Shyr C, Ren B, Hsu C-Y, et al. A statistical framework for evaluating the repeatability and reproducibility of large language models. medRxiv. Preprint. doi:10.1101/2025.08.06.25333170. First posted 2025; latest version posted March 25, 2026.

DOI: 10.1101/2025.08.06.25333170

HealthRadar editorial summary

This preprint proposes a regulatory-informed statistical framework for measuring large-language-model consistency along two dimensions: semantic consistency in the meaning of generated answers and internal consistency in the model’s token-level output distributions. The framework was applied across multiple models and five prompting approaches using 518 medical examination questions and 90 real-world rare-disease cases. Repeatability and reproducibility varied by model, dataset, and prompting strategy; prompts encouraging Bayesian diagnostic reasoning produced higher semantic repeatability in some comparisons. Most importantly, consistency was generally not associated with diagnostic accuracy. A model could therefore produce a correct answer inconsistently or repeat an incorrect answer reliably. The work argues that single-run accuracy testing is inadequate for evaluating stochastic medical systems and that correctness, repeatability, and cross-prompt reproducibility should be measured separately. This paper has not been peer reviewed.

Why it matters to HealthRadar

The framework directly supports repeated-run evaluation and separate scoring of output quality and consistency in the HealthRadar Yardstick study.

Read the original article

HealthRadar editorial summaries

Uncertainty, anxiety, and online health searching

Receiving laboratory results can create uncertainty even when the results do not require urgent action. Patients may seek additional information through portals, search engines, artificial intelligence systems, family members, or other sources. The studies below examine laboratory-result comprehension, health anxiety, online health-information seeking, and cyberchondria. They help define a behavioral context relevant to patient-facing laboratory communication but do not establish that HealthRadar reduces anxiety or inappropriate searching.

Uncertainty, anxiety, and online health searchingPublished · Peer reviewedOriginal research / patient survey · 2015

The Effects of Web-Based Patient Access to Laboratory Results in British Columbia: A Patient Survey on Comprehension and Anxiety

Authors: Geneviève Mák, Heather Smith Fowler, Chad Leaver, Simon Hagens, Jennifer Zelmer

Journal/source: J Med Internet Res.

Full citation: Mák G, Smith Fowler H, Leaver C, Hagens S, Zelmer J. The Effects of Web-Based Patient Access to Laboratory Results in British Columbia: A Patient Survey on Comprehension and Anxiety. J Med Internet Res. 2015;17(8):e191. doi:10.2196/jmir.4350.

DOI: 10.2196/jmir.4350

HealthRadar editorial summary

This exploratory study compared adults in British Columbia who accessed laboratory results online with a general-population comparison group without Web-based access. Among 2,047 online-service users and 1,245 comparison participants, online users reported receiving results much faster: approximately 84% received them within a few days compared with about 38% of the comparison group. However, self-reported comprehension was lower among online users, with about 76% saying they understood their results compared with about 85% in the comparison group. Overall reported anxiety did not differ significantly between groups. The findings illustrate an important distinction between rapid access and meaningful understanding: making laboratory data available sooner does not necessarily ensure that patients understand what the results mean.

Why it matters to HealthRadar

The study directly examines the gap between access to laboratory results and comprehension—the gap HealthRadar is intended to help address. It also cautions against assuming that providing results earlier necessarily produces either greater anxiety or greater understanding.

Read the original article

Uncertainty, anxiety, and online health searchingPublished · Peer reviewedSystematic review and meta-analysis · 2019

The relationships between health anxiety, online health information seeking, and cyberchondria: Systematic review and meta-analysis

Authors: Ryan D McMullan, David Berle, Sandra Arnáez, Vladan Starcevic

Journal/source: J Affect Disord.

Full citation: McMullan RD, Berle D, Arnáez S, Starcevic V. The relationships between health anxiety, online health information seeking, and cyberchondria: Systematic review and meta-analysis. J Affect Disord. 2019;245:270-278. doi:10.1016/j.jad.2018.11.037.

DOI: 10.1016/j.jad.2018.11.037

HealthRadar editorial summary

This systematic review and meta-analysis examined relationships among health anxiety, online health-information seeking, and cyberchondria. Twenty studies involving 7,373 participants were included across two meta-analyses. Health anxiety showed a moderate positive association with online health-information seeking (r=0.34; 95% CI, 0.20–0.48) and a substantially stronger association with cyberchondria (r=0.62; 95% CI, 0.52–0.71). The findings demonstrate association rather than causation: they do not establish that searching causes anxiety or that reducing searching reduces anxiety. They do, however, support the idea that repeated or distressing health-information seeking occurs within an important behavioral context of health uncertainty and health anxiety.

Why it matters to HealthRadar

Patients uncertain about laboratory findings may turn to online resources for explanation or reassurance. This study supports examining that behavior directly rather than assuming that more information seeking always resolves uncertainty.

Read the original article

Uncertainty, anxiety, and online health searchingPublished · Peer reviewedSystematic review and meta-analysis · 2021

Conceptualizations of Cyberchondria and Relations to the Anxiety Spectrum: Systematic Review and Meta-analysis

Authors: Sandra K Schenkel, Stefanie M Jungmann, Maria Gropalis, Michael Witthöft

Journal/source: J Med Internet Res.

Full citation: Schenkel SK, Jungmann SM, Gropalis M, Witthöft M. Conceptualizations of Cyberchondria and Relations to the Anxiety Spectrum: Systematic Review and Meta-analysis. J Med Internet Res. 2021;23(11):e27835. doi:10.2196/27835.

DOI: 10.2196/27835

HealthRadar editorial summary

This systematic review examined how cyberchondria has been conceptualized and how it relates to anxiety-spectrum phenomena. Twenty-five studies were included in the qualitative synthesis, and seven studies involving 3,069 participants contributed to the quantitative synthesis. Cyberchondria showed a strong association with health anxiety (r=0.63), while individual cyberchondria dimensions showed associations ranging from r=0.24 to r=0.66. The authors also identified relationships with intolerance of uncertainty, anxiety sensitivity, and obsessive-compulsive symptoms. They proposed that health-related internet searching in the context of health anxiety may function as a safety or reassurance-seeking behavior maintained by intermittent reinforcement. Because most underlying evidence is observational, the direction of causation remains uncertain.

Why it matters to HealthRadar

The review provides a behavioral model for why repeated searching may not always resolve health-related uncertainty. It supports studying whether clearer, bounded laboratory communication changes situational uncertainty or information-seeking behavior, without implying that HealthRadar is a treatment for anxiety.

Read the original article