Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care

Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care Michael Zuckerman , Gal Eyal , Roei Magen , Nir Lewis , Omer Harnof , Shiri Shifman , Zach Avraham , Lynn Joffe , Kevin Gallagher , View ORCID Profile Yair E. Lewis doi: https://doi.org/10.1101/2025.11.07.25339620 Michael Zuckerman 1 Navina AI , New York City, NY, United States MD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Gal Eyal 1 Navina AI , New York City, NY, United States MSc, MBA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Roei Magen 1 Navina AI , New York City, NY, United States MD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Nir Lewis 1 Navina AI , New York City, NY, United States MD, PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Omer Harnof 1 Navina AI , New York City, NY, United States MD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Shiri Shifman 1 Navina AI , New York City, NY, United States BSc Find this author on Google Scholar Find this author on PubMed Search for this author on this site Zach Avraham 1 Navina AI , New York City, NY, United States BSc Find this author on Google Scholar Find this author on PubMed Search for this author on this site Lynn Joffe 2 DTC Family Health , Denver, CO, United States MD, MSPH Find this author on Google Scholar Find this author on PubMed Search for this author on this site Kevin Gallagher 3 Hudson Headwaters Health Network , Queensbury, NY, United States MD Find this author on Google Scholar Find this author on PubMed Search for this author on this site Yair E. Lewis 1 Navina AI , New York City, NY, United States MD, PhD Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Yair E. Lewis For correspondence: lewis{at}navina.ai Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Background Ambient artificial intelligence (AI) clinical documentation tools have gained rapid adoption in healthcare to address physician burnout from documentation burden. However, current implementations primarily rely on real-time audio capture without systematically incorporating longitudinal patient data, potentially limiting documentation completeness for chronic disease management. Objective To compare documentation completeness between ambient audio-only workflows and those augmented with historical clinical data from electronic health records (EHRs) for type 2 diabetes and hypertension encounters in primary care. Methods We conducted a retrospective, paired, cross-sectional study of 354 primary care encounters in which diabetes mellitus (DM, n=119) and/or hypertension (HTN, n=281) were treated. Each condition instance was analysed twice to compare two methods of automated documentation: using only physician-patient conversation transcripts (termed “ambient only”) compared with consolidated automated documentation that includes historical clinical data in addition to the ambient conversation (ambient + history; termed “consolidated”). Documentation completeness was assessed using the “assessment” subset of the QNOTE clinical documentation quality measurement instrument, evaluating four domains: completeness, clinical coherence, clarity, conciseness. Scoring was automated using an LLM pipeline with physician validation on a 20% sample. Results Consolidated documentation achieved significantly higher mean total assessment composite score compared to ambient-only (94.8 vs. 80.1 on a scale of 0-100; difference 14.6 points; 95% CI 13.4-15.8; P<0.001). The largest improvements was observed in the completeness domain (difference 42.5 points; P<0.001). DM and HTN both showed similar performance of consolidated documentation vs. ambient only. Conclusions Augmenting ambient AI documentation with historical EHR data significantly improves documentation completeness for chronic disease management in primary care. These preliminary findings challenge the prevailing audio-first implementation paradigm and suggest that bidirectional EHR integration may be essential for comprehensive AI-assisted documentation, particularly for conditions requiring synthesis of longitudinal clinical data. Background Clinical documentation has emerged as a primary driver of physician burnout and dissatisfaction in modern healthcare. Studies have demonstrated that physicians spend a substantial amount of their workday engaging with electronic health records (EHRs) [ 1 ], with documentation consuming up to two hours for every hour of direct patient care[ 2 ]. This burden has been one of the drivers of the rapid adoption of ambient artificial intelligence (AI) clinical documentation tools, with rapid adoption across the US healthcare system[ 3 , 4 ]. Ambient clinical documentation leverages natural language processing and generative AI to passively capture patient-clinician conversations and automatically generate structured clinical notes. Healthcare organisations implementing these solutions report multiple benefits including decreased physician burnout, improved same-day closure rates, and enhanced clinical workflows. The evidence base for time savings is growing but heterogeneous. A University of Pennsylvania study found a 20.4% reduction in time spent per appointment in notes (from 10.3 to 8.2 minutes)[ 5 ], while similar implementations at Sutter Health demonstrated a 14.5% decrease (from 6.2 to 5.3 minutes)[ 6 ]. Critically, these tools reduce after-hours documentation by 2.5 to 3 hours weekly, directly addressing “pajama time” that erodes work-life balance[ 5 ]. Beyond time metrics, ambient scribing impacts documentation quality and completeness, though results vary considerably. Early evaluations suggest that AI-generated notes capture more detailed histories of present illness and may include clinical information that physicians might otherwise omit for brevity[ 7 ]. However, ambient scribing systems face fundamental limitations in capturing the full clinical encounter; they cannot document physical examination findings unless explicitly verbalised by the clinician, miss non-verbal cues such as patient affect or subtle physical signs, and fail to capture clinical observations that physicians naturally observe but may not articulate during the visit. This reliance on spoken content alone creates systematic gaps in documentation that are particularly problematic for comprehensive chronic disease management. The requirement for diligent clinician oversight introduces “editing fatigue” and automation bias risks, potentially undermining efficiency gains[ 8 ]. This reliance on audio-only workflows represents a critical gap in chronic disease management, where comprehensive documentation requires synthesis of current observations with historical trends, laboratory values, and treatment responses. For conditions like diabetes and hypertension, quality metrics demand documentation of glycemic control trajectories, medication adjustments, screening for complications, and preventive care measures-elements that may not surface naturally in routine conversations. In addition, incomplete documentation can impact provider compensation. The Medicare Evaluation and Management (E/M) documentation requirements, particularly the MEAT (Monitoring, Evaluation, Assessment, Treatment) criteria for chronic conditions, necessitate explicit linkage between current findings and historical data to support medical necessity and care continuity. Recent technical developments suggest that integrating historical context is feasible, with studies demonstrating that large language models can effectively synthesise multiple data sources for clinical documentation. However, systematic evaluation of how historical data augmentation affects documentation completeness, particularly for chronic disease management, remains unexplored in the peer-reviewed literature. This study addresses this gap by directly comparing documentation completeness between ambient audio-only workflows and those augmented with historical clinical data from EHRs and health information exchanges (HIE). By utilising a subset of the QNOTE clinical documentation quality scoring framework [ 9 ], and applying it to diabetes and hypertension encounters, we provide the first empirical assessment of whether incorporating longitudinal patient context enhances the quality of AI-generated clinical documentation. This evaluation is particularly timely given mixed evidence on time savings and growing recognition that persistent chart review needs may offset efficiency gains when critical historical information is omitted from ambient-only documentation. Methods Study Design We conducted a retrospective, paired, cross-sectional study to evaluate the completeness of primary care documentation generated through two methods. Each clinical encounter was documented twice: once using only the physician–patient conversation transcript (ambient-only), and once augmented with structured data from the electronic health record (ambient+history; “consolidated”). We compared the two approaches using a subset of the validated QNOTE scoring framework [ 9 ] ( Figure 1 ). Download figure Open in new tab Figure 1. Study Design Setting and Participants The study included de-identified AI generated ambient clinical scribing notes of primary care encounters from two primary care clinics in the United States. We included encounters in which for type 2 diabetes mellitus and hypertension were treated, selecting based on the ICD-10 codes that were coded in the encounter. Only the “assessment and plan” section of the encounter was evaluated, as per study goal. All encounters meeting the pre-specified inclusion criteria during the study period were included; no sampling or additional selection was undertaken. Although visits often addressed multiple conditions, we analysed only documentation elements related to those two conditions mentioned above, as a proxy for chronic disease documentation. Documentation Generation Ambient-only documentation was generated by an ambient clinical documentation platform (Navina, NY, USA) that produces structured SOAP notes directly from the ambient patient-physician interaction. As mentioned above, only the assessment and plan section was used in this study. Consolidated documentation was created by an combining the ambient documentation with longitudinal patient data from the electronic health record (EHR), including medications, laboratory values, family history, and consultation notes. The input was processed through an automated retrieval-augmented generation (RAG) pipeline which integrated the conversational and historical clinical data into a unified clinical note. The selection of relevant clinical data elements for each condition was based on a knowledge graph manually curated by physician domain experts, which enumerated condition-to-evidence linkages (e.g., key labs, medications, problem lists, imaging, and consult notes) used by the retriever to filter and rank candidate facts. The foundation model used in creating the consolidated note was Claude Sonnet 4 (Anthropic, CA, USA). Scoring Framework Documentation quality was assessed using a modified version of the QNOTE instrument, a validated tool for evaluating the quality of clinical notes in electronic health records [ 9 ]. Specifically, we adapted Section 10 (Assessment) of the QNOTE framework, which focuses on the quality of assessment section of the encounter documentation, across four domains ( Table 1 ). We operationalised this section into 14 sub-questions. Each sub question in the instrument was scored according to the following scale: “Full” (100 points); “Partial” (50 points); or “Unacceptable” (0 points). Domain scores were calculated as the arithmetic mean of all items within each respective domain, regardless of the number of items per domain. The overall composite score was derived using an equal-weighting approach across all four domains. Specifically, each domain contributed 25% to the final composite score, calculated as the unweighted mean of the four domain scores. View this table: View inline View popup Download powerpoint Table 1. QNOTE Instrument - Assessment Section Full credit is achievable even when elements are not applicable (e.g., documenting “no complications” or “not taking any medications”). Scoring was done via a Python pipeline, utilising GPT-5 (OpenAI, CA, USA) to score the modified QNOTE framework as described above. Scoring Validation To validate scoring reliability, two medical subject-matter experts rated the same random sample of notes (n = 20) using the modified QNOTE instrument described above. Both experts scored each note independently and in a blinded fashion, without knowledge of whether the note originated from the ambient-only or consolidated version. Agreement was assessed both between human raters (inter-reviewer agreement) and between the human consensus score and the LLM-generated score (model– human agreement), to confirm the validity and reproducibility of the automated evaluation. We calculated agreement at two thresholds: exact agreement (identical scores) and substantial agreement (score differences <50 points). Inter-rater exact agreement was 78.4% and substantial agreement was 95.3%. Human-model exact agreement was 73.8% and substantial agreement was 91.9%. Statistical Analysis All statistical analyses were performed using R (version 4.3.2; R Core Team, 2023). Given our paired cross-sectional design where each condition instance was analysed using both documentation methods, we employed paired Student’s t-tests to compare QNOTE assessment scores between ambient-only and consolidated (ambient + history) documentation. Descriptive statistics were calculated for both documentation methods across all four QNOTE assessment domains (completeness, clinical coherence, clarity, and conciseness) as well as the total assessment composite score. For each domain score and the composite score, 95% confidence intervals were computed for mean estimates within each documentation arm. Analyses were stratified by condition and conducted at three levels: (1) diabetes mellitus encounters (n=119), (2) hypertension encounters (n=281), and (3) the combined dataset. Given that some encounters included treatment for both conditions, each condition instance was analysed independently according to our paired design. To account for multiple comparisons across the four domains and composite score within each analysis stratum, we applied the Benjamini-Hochberg false discovery rate (FDR) correction. The statistical significance threshold was set at p < 0.05 (two-tailed). Results Overall score We evaluated 354 primary care encounters documenting management of type 2 diabetes mellitus (DM, n=119 condition instances) and essential hypertension (HTN, n=281 condition instances). Some encounters addressed both conditions. Each condition instance was independently documented using both methods (ambient-only and consolidated), creating paired comparisons for analysis. The overall mean assessment score was substantially higher for consolidated documentation than for ambient-only documentation ( Table 2 and Figure 2 ; 94.8 vs. 80.1; difference, 14.6 points; 95% CI,13.4 to 15.8; adjusted P<0.001). Statistically significant improvements were observed across all domains. The most significant gains occurred in completeness (difference, 42.5 points; adjusted P<0.001). Results were consistent across analysis of both conditions, DM and HTN ( Table 2 ). View this table: View inline View popup Download powerpoint Table 2. Results: Overall Assessment Score View this table: View inline View popup Download powerpoint Table 3. Results: Sub-Domain Level Download figure Open in new tab Figure 2. Results - Overall Assessment Score Nearly all encounter pairs showed higher scores with historical clinical data augmentation, indicating a consistent benefit of multimodal documentation. Subgroup analysis at the sub-domain level We analysed the results at the sub-domain level as well ( Table 3 and Figure 3 ). The largest differences were observed for vital signs (q5; 86.6 vs. 14.9; difference, 86.6 points), comorbidities (q8; 73.0 vs. 12.0; difference, 61.0 points), and past medical history (q6; 59.9 vs. 8.2; difference, 51.6 points). Core disease markers (q2; 97.8 vs. 42.4; difference, 55.4 points) and disease severity (q1; 100.0 vs. 56.8; difference, 43.2 points) also demonstrated marked improvements. Of note, there was marked variability in documentation quality for ambient-only, as opposed to much narrower distribution in the consolidated note group. This likely reflects variance in amount of conversation between encounters. Download figure Open in new tab Figure 3. Results - Sub-Domain Level Discussion This pilot study provides initial evidence suggesting that augmenting ambient AI documentation with historical clinical data may improve documentation completeness for chronic disease management in primary care. The 14.6-point improvement in overall assessment score represents a potentially meaningful enhancement, particularly given the 42.5-point gain in the “completeness” domain - a critical aspect of quality of clinical documentation. Our findings relate to a fundamental limitation of audio-only ambient documentation: the verbal patient-physician interaction may omit quantitative and historical information essential for comprehensive chronic care. The substantial differences between ambient-only and consolidated notes in the completeness domains of core disease markers, vital signs, and condition relevant comorbidities demonstrate that critical elements like laboratory trends, vital sign patterns, and complication screening results rarely surface naturally in clinical conversations. The minimal impact on medication plan and clinical reasoning documentation suggests that treatment discussions are already well-captured through conversation alone. This finding aligns with workflow observations that medication changes and care planning constitute major components of verbal physician-patient interactions. These preliminary results have important implications for ambient AI implementation strategies. While current deployments emphasize real-time transcription efficiency, our data suggest that organisations should prioritise bidirectional EHR integration, i.e. the ability to retrieve historical clinical data, to realise the full potential of AI documentation assistance. The study also illuminates the tension between documentation efficiency and completeness. Although ambient-only tools demonstrably reduce documentation time, the resulting notes may require substantial post-encounter chart review and editing to meet quality and reimbursement standards - potentially offsetting some of the initial time savings. Historical data integration could improve efficiency by reducing the need for manual note augmentation, while ensuring compliance for E/M billing. Limitations Several limitations warrant consideration. First, this pilot analysed a small dataset of encounters by a limited number of physicians, in two clinical sites, limiting generalisability. Second, we evaluated completeness rather than accuracy or clinical relevance. Finally, our analysis focused exclusively on diabetes and hypertension; conditions with different documentation requirements may show varying benefit patterns. Future Directions This pilot study justifies more rigorous investigation into optimal AI documentation strategies. Future research using a larger and more diverse dataset should evaluate real-world outcomes including physician time savings when accounting for note review and editing, clinical decision-making impact, billing accuracy, and patient safety considerations. Investigating optimal strategies for presenting historical context - balancing comprehensiveness with cognitive load - will be crucial for successful implementation. As ambient AI tools rapidly proliferate across healthcare systems, these preliminary findings underscore the importance of moving beyond simple transcription and toward intelligent documentation systems that synthesise conversational and historical, longitudinal, clinical data. While confirmatory studies are needed, this initial evidence suggests that integrating historical context may support more comprehensive documentation for chronic disease management in primary care. Data Availability Data produced in the present study are available upon reasonable request to the authors, other than the de-identified clinical notes which cannot be shared in order to protect patient privacy and in accordance with HIPAA regulations. Conflict of Interest Disclosures MZ, GE, RM, NL, OH, SS, ZA and YEL are employees of Navina, which produces AI-enabled clinical assistant tools for physicians. Ethics Statement All data used in the study was fully de-identified prior to analysis. Study was determined as non-human subject research (NSHR) by ethics committees of both clinical sites (DTC, HHHN). Table 3. Results: Sub-Domain Level Figure 3 : Results: Sub-Domain Level Bibliography 1. ↵ Overhage , J. Marc , and David McCallie Jr . “ Physician time spent using the electronic health record during outpatient encounters: a descriptive study .” Annals of internal medicine 172 . 3 ( 2020 ): 169 – 174 . OpenUrl CrossRef PubMed 2. ↵ Sinsky , Christine , et al. “ Allocation of physician time in ambulatory practice: a time and motion study in 4 specialties .” Annals of internal medicine 165 . 11 ( 2016 ): 753 – 760 . OpenUrl CrossRef PubMed 3. ↵ Tierney , Aaron A. , et al. “ Ambient artificial intelligence scribes: learnings after 1 year and over 2.5 million uses .” NEJM Catalyst Innovations in Care Delivery 6 . 5 ( 2025 ): CAT – 25 . OpenUrl 4. ↵ Poon , Eric G. , et al. “ Adoption of artificial intelligence in healthcare: survey of health system priorities, successes, and challenges .” Journal of the American Medical Informatics Association 32 . 7 ( 2025 ): 1093 – 1100 . OpenUrl CrossRef PubMed 5. ↵ Duggan , Matthew J. , et al. “ Clinician experiences with ambient scribe technology to assist with documentation burden and efficiency .” JAMA Network Open 8 . 2 ( 2025 ): e2460637 – e2460637 . OpenUrl 6. ↵ Stults , Cheryl D. , et al. “ Evaluation of an ambient artificial intelligence documentation platform for clinicians .” JAMA Network Open 8 . 5 ( 2025 ): e258614 – e258614 . OpenUrl PubMed 7. ↵ Albrecht , Michael , et al. “ Enhancing clinical documentation with ambient artificial intelligence: a quality improvement survey assessing clinician perspectives on work burden, burnout, and job satisfaction .” JAMIA open 8 . 1 ( 2024 ). 8. ↵ Leung , Tiffany I. , Andrew J. Coristine, and Arriel Benis. “AI Scribes in Health Care: Balancing Transformative Potential With Responsible Integration.” JMIR Medical Informatics 13 . 1 ( 2025 ): e80898 . OpenUrl 9. ↵ Burke , Harry B. , et al. “ QNOTE: an instrument for measuring the quality of EHR clinical notes .” Journal of the American Medical Informatics Association 21 . 5 ( 2014 ): 910 – 916 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted November 09, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care Michael Zuckerman , Gal Eyal , Roei Magen , Nir Lewis , Omer Harnof , Shiri Shifman , Zach Avraham , Lynn Joffe , Kevin Gallagher , Yair E. Lewis medRxiv 2025.11.07.25339620; doi: https://doi.org/10.1101/2025.11.07.25339620 Share This Article: Copy Citation Tools Ambient Only vs. Longitudinal Data-Enhanced AI Documentation: A Pilot Study Quantifying the Value of Historical Clinical Context in Primary Care Michael Zuckerman , Gal Eyal , Roei Magen , Nir Lewis , Omer Harnof , Shiri Shifman , Zach Avraham , Lynn Joffe , Kevin Gallagher , Yair E. Lewis medRxiv 2025.11.07.25339620; doi: https://doi.org/10.1101/2025.11.07.25339620 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Health Informatics Subject Areas All Articles Addiction Medicine (570) Allergy and Immunology (864) Anesthesia (302) Cardiovascular Medicine (4445) Dentistry and Oral Medicine (444) Dermatology (383) Emergency Medicine (609) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1515) Epidemiology (15236) Forensic Medicine (30) Gastroenterology (1127) Genetic and Genomic Medicine (6610) Geriatric Medicine (669) Health Economics (1000) Health Informatics (4549) Health Policy (1370) Health Systems and Quality Improvement (1613) Hematology (543) HIV/AIDS (1266) Infectious Diseases (except HIV/AIDS) (15926) Intensive Care and Critical Care Medicine (1104) Medical Education (623) Medical Ethics (147) Nephrology (668) Neurology (6613) Nursing (346) Nutrition (999) Obstetrics and Gynecology (1147) Occupational and Environmental Health (957) Oncology (3341) Ophthalmology (975) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (665) Pediatrics (1694) Pharmacology and Therapeutics (693) Primary Care Research (714) Psychiatry and Clinical Psychology (5458) Public and Global Health (9244) Radiology and Imaging (2205) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1197) Rheumatology (596) Sexual and Reproductive Health (715) Sports Medicine (530) Surgery (713) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a0259ed71b01dd69',t:'MTc3OTg5MTIwOA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();