Ata M. Kiapour
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ata M. Kiapour has a total of 225 co-authors. Here are the top co-authors by number of shared publications:
- Braden C. Fleming
Department of Orthopaedics, Warren Alpert Medical School of Brown University, Rhode Island Hospital, Providence, Rhode Island, USA
20 shared publications Kirsten Ecklund
Department of Radiology, Boston Children's Hospital, Harvard Medical School, Boston, Massachusetts, USA
12 shared publications- Martha Meaney Murray
Boston Children’s Hospital, Harvard Medical School
12 shared publications - Eduardo N. Novais
Department of Orthopedic Surgery Boston Children's Hospital, Boston, MA, USA
11 shared publications Martha M. Murray
Department of Orthopaedic Surgery and Sports Medicine, Boston Children’s Hospital, Harvard Medical School, Boston, MA
11 shared publications- Nazgol Tavabi
Department of Orthopedics and Sports Medicine, Boston Children's Hospital, Harvard Medical School, Boston, MA 02115, United States
11 shared publications
Latest Publications
- Open Access
Agentic Pipeline for Large-Scale Clinical Data Queries in Enterprise Electronic Health Records
Using electronic health record (EHR) data to answer clinical research questions is challenging. It requires understanding the research question, finding relevant information across different sources, and combining that information to produce meaningful results. We developed an agentic large language model (LLM) pipeline that decomposes investigator-authored requests into source-specific tasks, generates and executes SQL on EHR, integrates structured and unstructured EHR evidence, and applies study-specific clinical note adjudication. We evaluated this workflow against data analyst–identified cohorts across 50 pediatric orthopedic clinical data query requests. Of these, 43 were included in manual-review-adjusted performance analyses. The median adjusted positive predictive value, sensitivity among patients identified by either approach, and corresponding F1 score were 0.93 (95% CI, 0.88–0.98), 0.92 (95% CI, 0.90–0.95) and 0.90 (95% CI, 0.87–0.94), respectively. These findings support the feasibility of translating natural-language research intent into executable, autonomous multi-stage EHR workflows.
Open AccessAgentic Pipeline for Large-Scale Clinical Data Queries in Enterprise Electronic Health Records
Using electronic health record (EHR) data to answer clinical research questions is challenging. It requires understanding the research question, finding relevant information across different sources, and combining that information to produce meaningful results. We developed an agentic large language model (LLM) pipeline that decomposes investigator-authored requests into source-specific tasks, generates and executes SQL on EHR, integrates structured and unstructured EHR evidence, and applies study-specific clinical note adjudication. We evaluated this workflow against data analyst–identified cohorts across 50 pediatric orthopedic clinical data query requests. Of these, 43 were included in manual-review-adjusted performance analyses. The median adjusted positive predictive value, sensitivity among patients identified by either approach, and corresponding F1 score were 0.93 (95% CI, 0.88–0.98), 0.92 (95% CI, 0.90–0.95) and 0.90 (95% CI, 0.87–0.94), respectively. These findings support the feasibility of translating natural-language research intent into executable, autonomous multi-stage EHR workflows.
- Open Access
- 11 August 2026
- Published by SAGE Publications in The American Journal of Sports Medicine
- Vol. 54 (11) , 2697-2704
Longitudinal Quantitative Mapping of Meniscal Morphological and Signal Intensity Differences After ACL Surgery
Background: Meniscal tears are frequently associated with anterior cruciate ligament (ACL) injuries. Evaluating longitudinal changes in meniscal integrity using quantitative magnetic resonance imaging after ACL surgery may provide valuable insight into joint recovery and subsequent reinjury risk. Hypotheses: (1) The cross-sectional area (CSA) and normalized signal intensity (SI) profiles of the lateral and medial menisci for the surgical knees would be greater than those of the contralateral uninjured knees observed at 6, 12, and 24 months after surgery. (2) The CSA and Constructive Interference in Steady State (CISS) SI profiles obtained at 24 months would be closer to those of the contralateral limb when compared with those of the 6- and 12-month time points. Study Design: Cohort study; Level of evidence, 3. Methods: Deidentified postsurgical CISS magnetic resonance imaging scans of patients (n = 82; median age 17.6 years; 42 females and 40 males; 57 ACL restoration and 25 ACL reconstruction) were analyzed. CSA and normalized CISS SI values were extracted from the lateral and medial menisci to create morphological profiles. Statistical parametric mapping was used to compare the CSA and normalized CISS SI profiles between the surgical and contralateral limbs to identify specific regions that were different among the menisci at 6, 12, and 24 months. Results: The CSA and normalized CISS SI profiles of the anterior region of the medial menisci and lateral menisci for the surgical knees were greater than those of the contralateral uninjured knees, observed at 6, 12, and 24 months after surgery, whereas the posterior region CSA of the medial menisci was smaller. The lateral meniscus on the operated limb had a larger CSA and higher normalized CISS SI values than the lateral meniscus in the contralateral limb. The CSA and CISS SI profiles obtained at 24 months did not approach those of the contralateral limb. Conclusion: Significant changes occurred in both menisci after ACL surgery that did not return to baseline levels within 2 years. It is important to account for anatomic location when evaluating meniscal morphology in surgical knees and when evaluating the effects of ACL injuries on the surrounding tissues.
- 11 August 2026
- Published by SAGE Publications in The American Journal of Sports Medicine
- Vol. 54 (11) , 2697-2704
Open AccessLongitudinal Quantitative Mapping of Meniscal Morphological and Signal Intensity Differences After ACL Surgery
Background: Meniscal tears are frequently associated with anterior cruciate ligament (ACL) injuries. Evaluating longitudinal changes in meniscal integrity using quantitative magnetic resonance imaging after ACL surgery may provide valuable insight into joint recovery and subsequent reinjury risk. Hypotheses: (1) The cross-sectional area (CSA) and normalized signal intensity (SI) profiles of the lateral and medial menisci for the surgical knees would be greater than those of the contralateral uninjured knees observed at 6, 12, and 24 months after surgery. (2) The CSA and Constructive Interference in Steady State (CISS) SI profiles obtained at 24 months would be closer to those of the contralateral limb when compared with those of the 6- and 12-month time points. Study Design: Cohort study; Level of evidence, 3. Methods: Deidentified postsurgical CISS magnetic resonance imaging scans of patients (n = 82; median age 17.6 years; 42 females and 40 males; 57 ACL restoration and 25 ACL reconstruction) were analyzed. CSA and normalized CISS SI values were extracted from the lateral and medial menisci to create morphological profiles. Statistical parametric mapping was used to compare the CSA and normalized CISS SI profiles between the surgical and contralateral limbs to identify specific regions that were different among the menisci at 6, 12, and 24 months. Results: The CSA and normalized CISS SI profiles of the anterior region of the medial menisci and lateral menisci for the surgical knees were greater than those of the contralateral uninjured knees, observed at 6, 12, and 24 months after surgery, whereas the posterior region CSA of the medial menisci was smaller. The lateral meniscus on the operated limb had a larger CSA and higher normalized CISS SI values than the lateral meniscus in the contralateral limb. The CSA and CISS SI profiles obtained at 24 months did not approach those of the contralateral limb. Conclusion: Significant changes occurred in both menisci after ACL surgery that did not return to baseline levels within 2 years. It is important to account for anatomic location when evaluating meniscal morphology in surgical knees and when evaluating the effects of ACL injuries on the surrounding tissues.
- Open Access
- 10 June 2026
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 14 (6)
Magnetic Resonance Imaging Assessment of Bridge-Enhanced ACL Restoration: An Imaging Atlas of Morphological Features and Healing Trajectories
Background: The use of bridge-enhanced anterior cruciate ligament (ACL) restoration (BEAR) procedures is on the rise. Yet, there is a paucity of literature describing the radiographic appearance of the ACL after this procedure. Purpose: To present the quantitative and qualitative postoperative changes in healing ACLs within the first 24 months after an ACL restoration procedure. Study Design: Case series; Level of evidence, 4. Methods: A total of 64 patients (64 knees, 19.4 ± 5 years, 57% women) underwent magnetic resonance imaging (MRI) of the knee 6, 12, and 24 months after ACL restoration surgery with an extracellular matrix scaffold as part of a randomized-controlled trial. The ACL sagittal elevation angle and tunnel positions, along with ACL volume, mean cross-sectional area (CSA), length, and normalized signal intensity (SI), were quantified from MRI. Patients were divided into quartiles based on SI at 24 months after surgery for inclusion in the qualitative analysis with an adapted framework for assessing tissue quality. Additional cases were selected for qualitative analysis based on clinical representations. Linear mixed modeling with Tukey post hoc corrections was used to analyze quantitative ACL healing parameters. Results: Tunnel positions were anatomic in 26.6% of femurs and 68.8% of tibias. The healing ACL sagittal elevation angle remained stable ( .1) and was comparable with the contralateral native ACL ( .05) at all time points. The healing ACL volume and CSA decreased over time ( P < .01) but remained 20% (volume) and 15% (CSA) larger than the contralateral ACL at 24 months postoperatively (by 341 mm 3 volume and 6 mm 2 CSA; P < .001). The healing ACL length was constant over time ( .1) and was 5% longer than the contralateral native ACL at final follow-up (by 1.8 mm; P = .027). The ACL SI decreased over time ( P < .001) and was 10% less than the contralateral ACL at 24 months (by 0.1 ± 0.3; P = .022). Seven relevant clinical scenarios were highlighted to show a spectrum of healing trajectories after ACL restoration surgery. Conclusion: The present study demonstrates that throughout 2 years of serial imaging after an ACL restoration procedure, the healing ACL becomes smaller cross-sectionally and more organized (lower normalized SI). Despite changes in the ACL size and ACL SI, the sagittal orientation remains unchanged and comparable with the contralateral native ACL. In the qualitative analysis, 7 distinct healing trajectories were identified. These data are important for interpreting postoperative imaging after the BEAR procedure, and for identifying that several healing trajectories exist after a restoration procedure. Further studies are warranted to identify patient-specific factors that may influence these trajectories. Registration: NCT02664545 (ClinicalTrials.gov identifier).
- 10 June 2026
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 14 (6)
Open AccessMagnetic Resonance Imaging Assessment of Bridge-Enhanced ACL Restoration: An Imaging Atlas of Morphological Features and Healing Trajectories
Background: The use of bridge-enhanced anterior cruciate ligament (ACL) restoration (BEAR) procedures is on the rise. Yet, there is a paucity of literature describing the radiographic appearance of the ACL after this procedure. Purpose: To present the quantitative and qualitative postoperative changes in healing ACLs within the first 24 months after an ACL restoration procedure. Study Design: Case series; Level of evidence, 4. Methods: A total of 64 patients (64 knees, 19.4 ± 5 years, 57% women) underwent magnetic resonance imaging (MRI) of the knee 6, 12, and 24 months after ACL restoration surgery with an extracellular matrix scaffold as part of a randomized-controlled trial. The ACL sagittal elevation angle and tunnel positions, along with ACL volume, mean cross-sectional area (CSA), length, and normalized signal intensity (SI), were quantified from MRI. Patients were divided into quartiles based on SI at 24 months after surgery for inclusion in the qualitative analysis with an adapted framework for assessing tissue quality. Additional cases were selected for qualitative analysis based on clinical representations. Linear mixed modeling with Tukey post hoc corrections was used to analyze quantitative ACL healing parameters. Results: Tunnel positions were anatomic in 26.6% of femurs and 68.8% of tibias. The healing ACL sagittal elevation angle remained stable ( .1) and was comparable with the contralateral native ACL ( .05) at all time points. The healing ACL volume and CSA decreased over time ( P < .01) but remained 20% (volume) and 15% (CSA) larger than the contralateral ACL at 24 months postoperatively (by 341 mm 3 volume and 6 mm 2 CSA; P < .001). The healing ACL length was constant over time ( .1) and was 5% longer than the contralateral native ACL at final follow-up (by 1.8 mm; P = .027). The ACL SI decreased over time ( P < .001) and was 10% less than the contralateral ACL at 24 months (by 0.1 ± 0.3; P = .022). Seven relevant clinical scenarios were highlighted to show a spectrum of healing trajectories after ACL restoration surgery. Conclusion: The present study demonstrates that throughout 2 years of serial imaging after an ACL restoration procedure, the healing ACL becomes smaller cross-sectionally and more organized (lower normalized SI). Despite changes in the ACL size and ACL SI, the sagittal orientation remains unchanged and comparable with the contralateral native ACL. In the qualitative analysis, 7 distinct healing trajectories were identified. These data are important for interpreting postoperative imaging after the BEAR procedure, and for identifying that several healing trajectories exist after a restoration procedure. Further studies are warranted to identify patient-specific factors that may influence these trajectories. Registration: NCT02664545 (ClinicalTrials.gov identifier).
- Open Access
- 21 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Large language models for zero-shot procedure extraction in orthopedic surgery: a comparative evaluation
Operative notes in electronic health records contain critical information for understanding surgical care, yet manual coding is time-consuming, costly, and inconsistent. Large language models (LLMs) promise to transform this process by automatically extracting detailed procedure information - a capability with significant implications for scaling clinical registries and advancing surgical research. Here, we conducted a large-scale evaluation of state-of-the-art LLMs for zero-shot structured information extraction from orthopedic clinical notes. Fourteen open-source and proprietary models were tested on 800 real operative notes, annotated by both an orthopedic surgeon and an administrator using a curated list of 74 procedure classes. We compared model outputs to human annotations, assessing accuracy and exploring the effects of model scale, reasoning capabilities, and prompt design. We find that across models, LLMs consistently outperform administrator-assigned labels, achieving macro-F1 scores above 0.6 and improving over administrative coding by up to 10 points. Larger models and reasoning capabilities further boosted performance, though gains plateaued beyond 30 billion parameters. Performance varied by procedure frequency, revealing clear strengths and persistent challenges for rare or complex cases. Modern LLMs can already outperform routine administrative coding in extracting detailed surgical procedure data, pointing to a future where registry curation could be faster, cheaper, and more consistent. Yet, full alignment with surgical experts remains an open challenge- especially for rare procedures - emphasizing the need for domain adaptation and thoughtful deployment. Our findings illustrate how general-purpose LLMs can advance automated clinical data curation and inform the next generation of surgical informatics.
- 21 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessLarge language models for zero-shot procedure extraction in orthopedic surgery: a comparative evaluation
Operative notes in electronic health records contain critical information for understanding surgical care, yet manual coding is time-consuming, costly, and inconsistent. Large language models (LLMs) promise to transform this process by automatically extracting detailed procedure information - a capability with significant implications for scaling clinical registries and advancing surgical research. Here, we conducted a large-scale evaluation of state-of-the-art LLMs for zero-shot structured information extraction from orthopedic clinical notes. Fourteen open-source and proprietary models were tested on 800 real operative notes, annotated by both an orthopedic surgeon and an administrator using a curated list of 74 procedure classes. We compared model outputs to human annotations, assessing accuracy and exploring the effects of model scale, reasoning capabilities, and prompt design. We find that across models, LLMs consistently outperform administrator-assigned labels, achieving macro-F1 scores above 0.6 and improving over administrative coding by up to 10 points. Larger models and reasoning capabilities further boosted performance, though gains plateaued beyond 30 billion parameters. Performance varied by procedure frequency, revealing clear strengths and persistent challenges for rare or complex cases. Modern LLMs can already outperform routine administrative coding in extracting detailed surgical procedure data, pointing to a future where registry curation could be faster, cheaper, and more consistent. Yet, full alignment with surgical experts remains an open challenge- especially for rare procedures - emphasizing the need for domain adaptation and thoughtful deployment. Our findings illustrate how general-purpose LLMs can advance automated clinical data curation and inform the next generation of surgical informatics.
- Open Access
- 4 February 2026
- Published by Oxford University Press (OUP) in Journal of Hip Preservation Surgery
Ethnic differences in femoral and acetabular version in developmental dysplasia of the hip: a 3D CT-based comparative study of Japanese and Caucasian American cohorts
Developmental dysplasia of the hip (DDH) is associated with altered acetabular and femoral morphology that can influence hip joint mechanics and surgical planning. Although ethnic differences in hip morphology have been reported, no study has directly compared femoral version (FV), acetabular version (AV), combined version (CV), or FV–AV distribution patterns among DDH patients of different ethnic backgrounds. This retrospective study compared FV, AV, and their combined distribution between 55 hips from Japanese patients and 55 hips from Caucasian American patients with symptomatic DDH who underwent periacetabular osteotomy. Age and sex were matched using propensity score methods. FV and AV were classified as increased (>25°), normal (10–25°), or decreased (<10°). Between-group differences were assessed for FV, AV, CV, and FV–AV combinations. Japanese patients showed significantly greater FV (mean difference + 5.5°, P < .05) and CV (mean difference + 5.5°, P < .05) compared with Caucasian American patients. FV–AV distribution patterns differed significantly between groups (P = .0031). The most frequent pattern in Japanese patients was increased FV with normal AV (45.5%), whereas Caucasian American patients most commonly demonstrated normal FV with normal AV (36.4%). Decreased FV was more common in the Caucasian American cohort (10.9% vs. 5.5%). Ethnic differences were observed in the distribution patterns of FV and AV, suggesting morphological diversity in DDH that may influence its natural history. At the same time, these findings underscore the need for individualized assessment of hip morphology—rather than reliance on ethnic averages—when planning surgery and evaluating outcomes.
- 4 February 2026
- Published by Oxford University Press (OUP) in Journal of Hip Preservation Surgery
Open AccessEthnic differences in femoral and acetabular version in developmental dysplasia of the hip: a 3D CT-based comparative study of Japanese and Caucasian American cohorts
Developmental dysplasia of the hip (DDH) is associated with altered acetabular and femoral morphology that can influence hip joint mechanics and surgical planning. Although ethnic differences in hip morphology have been reported, no study has directly compared femoral version (FV), acetabular version (AV), combined version (CV), or FV–AV distribution patterns among DDH patients of different ethnic backgrounds. This retrospective study compared FV, AV, and their combined distribution between 55 hips from Japanese patients and 55 hips from Caucasian American patients with symptomatic DDH who underwent periacetabular osteotomy. Age and sex were matched using propensity score methods. FV and AV were classified as increased (>25°), normal (10–25°), or decreased (<10°). Between-group differences were assessed for FV, AV, CV, and FV–AV combinations. Japanese patients showed significantly greater FV (mean difference + 5.5°, P < .05) and CV (mean difference + 5.5°, P < .05) compared with Caucasian American patients. FV–AV distribution patterns differed significantly between groups (P = .0031). The most frequent pattern in Japanese patients was increased FV with normal AV (45.5%), whereas Caucasian American patients most commonly demonstrated normal FV with normal AV (36.4%). Decreased FV was more common in the Caucasian American cohort (10.9% vs. 5.5%). Ethnic differences were observed in the distribution patterns of FV and AV, suggesting morphological diversity in DDH that may influence its natural history. At the same time, these findings underscore the need for individualized assessment of hip morphology—rather than reliance on ethnic averages—when planning surgery and evaluating outcomes.
- Open Access
- 20 November 2025
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 13 (11)
Sagittal Orientation of the Acetabulum and Its Relationship to Spinopelvic Alignment: Three-Dimensional Assessment of 3700 Individuals
Background: The orientation of the acetabulum in the axial and coronal planes is well studied in the pathogenesis of impingement and instability of the hip. In contrast, the sagittal orientation of the acetabulum (SOA) is not well understood. Purpose: To determine (1) the SOA in a large cohort of mature hips and (2) to assess the relationship between the SOA and acetabular version, acetabular center-edge angles (CEAs), and spinopelvic alignment. Study Design: Descriptive laboratory study. Methods: A total of 3695 patients (7390 mature hips) who underwent computed tomography (CT) scans for assessment of nonorthopaedic abdominal and pelvic conditions were studied. An automated measurement software was utilized to reconstruct 3-dimensional models from CT scans and to measure the SOA, functional SOA (not neutralizing pelvic position on sagittal plane), acetabular version, as well as acetabular CEAs and spinopelvic alignment, including the pelvic tilt (PT), sacral slope (SS), and pelvic incidence (PI). Results: The SOA was on average (± SD) 19.6°± 7.5°. The functional SOA (not neutralizing pelvic position on sagittal plane) was on average (± SD) 20.5°± 5.7°. The functional SOA had a statistically significant but negligible correlation with PI ( r = 0.13; P < .001) and SS ( r = −0.06; P < .001), and a weak positive correlation with PT ( r = 0.23; P < .001). The SOA had a positive moderate correlation with the cranial ( r = 0.41; P < .001) and central acetabular version ( r = 0.39; P < .001) and a strong correlation ( r = 0.63; P < .001) with caudal acetabular version. A 10° increase in SOA was associated with a 6.6° increase on the caudal acetabular version. The SOA had a moderate negative correlation ( r = −0.48; P < .001) with the CEA at 3 o'clock (anterior for left and right hips). A 10° increase in SOA was associated with a 4.9° decrease in CEA at 3 o'clock. Conclusion: The acetabulum is on average 19.5° cephalically oriented in the sagittal plane in asymptomatic individuals. The SOA correlates with acetabular version and cannot be presumed based on spinopelvic alignment. Clinical Relevance: The assessment of the SOA may aid in the diagnosis of hip impingement and instability, allowing a more precise correction of the acetabulum in hip arthroscopy and osteotomies.
- 20 November 2025
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 13 (11)
Open AccessSagittal Orientation of the Acetabulum and Its Relationship to Spinopelvic Alignment: Three-Dimensional Assessment of 3700 Individuals
Background: The orientation of the acetabulum in the axial and coronal planes is well studied in the pathogenesis of impingement and instability of the hip. In contrast, the sagittal orientation of the acetabulum (SOA) is not well understood. Purpose: To determine (1) the SOA in a large cohort of mature hips and (2) to assess the relationship between the SOA and acetabular version, acetabular center-edge angles (CEAs), and spinopelvic alignment. Study Design: Descriptive laboratory study. Methods: A total of 3695 patients (7390 mature hips) who underwent computed tomography (CT) scans for assessment of nonorthopaedic abdominal and pelvic conditions were studied. An automated measurement software was utilized to reconstruct 3-dimensional models from CT scans and to measure the SOA, functional SOA (not neutralizing pelvic position on sagittal plane), acetabular version, as well as acetabular CEAs and spinopelvic alignment, including the pelvic tilt (PT), sacral slope (SS), and pelvic incidence (PI). Results: The SOA was on average (± SD) 19.6°± 7.5°. The functional SOA (not neutralizing pelvic position on sagittal plane) was on average (± SD) 20.5°± 5.7°. The functional SOA had a statistically significant but negligible correlation with PI ( r = 0.13; P < .001) and SS ( r = −0.06; P < .001), and a weak positive correlation with PT ( r = 0.23; P < .001). The SOA had a positive moderate correlation with the cranial ( r = 0.41; P < .001) and central acetabular version ( r = 0.39; P < .001) and a strong correlation ( r = 0.63; P < .001) with caudal acetabular version. A 10° increase in SOA was associated with a 6.6° increase on the caudal acetabular version. The SOA had a moderate negative correlation ( r = −0.48; P < .001) with the CEA at 3 o'clock (anterior for left and right hips). A 10° increase in SOA was associated with a 4.9° decrease in CEA at 3 o'clock. Conclusion: The acetabulum is on average 19.5° cephalically oriented in the sagittal plane in asymptomatic individuals. The SOA correlates with acetabular version and cannot be presumed based on spinopelvic alignment. Clinical Relevance: The assessment of the SOA may aid in the diagnosis of hip impingement and instability, allowing a more precise correction of the acetabulum in hip arthroscopy and osteotomies.
- Open Access
- 1 October 2025
- Published by Elsevier in Journal of the Pediatric Orthopaedic Society of North America
- Vol. 13, 100280
Obesity-associated Morphological Changes in Acetabular Growth and Development: Potential Implications for Slipped Capital Femoral Epiphysis
- 1 October 2025
- Published by Elsevier in Journal of the Pediatric Orthopaedic Society of North America
- Vol. 13, 100280
Open AccessObesity-associated Morphological Changes in Acetabular Growth and Development: Potential Implications for Slipped Capital Femoral Epiphysis
- Open Access
Large Language Models for Zero-Shot Procedure Extraction in Orthopedic Surgery: A Comparative Evaluation
Background Operative notes in electronic health records contain critical information for understanding surgical care, yet manual coding is time-consuming, costly, and inconsistent. Large language models (LLMs) promise to transform this process by automatically extracting detailed procedure information—a capability with significant implications for scaling clinical registries and advancing surgical research. Methods We conducted a large-scale evaluation of state-of-the-art LLMs for zero-shot structured information extraction from orthopedic clinical notes. Fourteen open-source and proprietary models were tested on 800 real operative notes, annotated by both an orthopedic surgeon and an administrator using a curated list of 74 procedure classes. We compared model outputs to human annotations, assessing accuracy and exploring the effects of model scale, reasoning capabilities, and prompt design. Results Across models, LLMs consistently outperformed administrator-assigned labels, achieving macro-F1 scores above 0.6 and improving over administrative coding by up to 10 points. Larger models and reasoning capabilities further boosted performance, though gains plateaued beyond 30 billion parameters. Performance varied by procedure frequency, revealing clear strengths and persistent challenges for rare or complex cases. Conclusion Modern LLMs can already outperform routine administrative coding in extracting detailed surgical procedure data, pointing to a future where registry curation could be faster, cheaper, and more consistent. Yet, full alignment with surgical experts remains an open challenge—especially for rare procedures—emphasizing the need for domain adaptation and thoughtful deployment. Our findings illustrate how general-purpose LLMs can advance automated clinical data curation and inform the next generation of surgical informatics.
Open AccessLarge Language Models for Zero-Shot Procedure Extraction in Orthopedic Surgery: A Comparative Evaluation
Background Operative notes in electronic health records contain critical information for understanding surgical care, yet manual coding is time-consuming, costly, and inconsistent. Large language models (LLMs) promise to transform this process by automatically extracting detailed procedure information—a capability with significant implications for scaling clinical registries and advancing surgical research. Methods We conducted a large-scale evaluation of state-of-the-art LLMs for zero-shot structured information extraction from orthopedic clinical notes. Fourteen open-source and proprietary models were tested on 800 real operative notes, annotated by both an orthopedic surgeon and an administrator using a curated list of 74 procedure classes. We compared model outputs to human annotations, assessing accuracy and exploring the effects of model scale, reasoning capabilities, and prompt design. Results Across models, LLMs consistently outperformed administrator-assigned labels, achieving macro-F1 scores above 0.6 and improving over administrative coding by up to 10 points. Larger models and reasoning capabilities further boosted performance, though gains plateaued beyond 30 billion parameters. Performance varied by procedure frequency, revealing clear strengths and persistent challenges for rare or complex cases. Conclusion Modern LLMs can already outperform routine administrative coding in extracting detailed surgical procedure data, pointing to a future where registry curation could be faster, cheaper, and more consistent. Yet, full alignment with surgical experts remains an open challenge—especially for rare procedures—emphasizing the need for domain adaptation and thoughtful deployment. Our findings illustrate how general-purpose LLMs can advance automated clinical data curation and inform the next generation of surgical informatics.
- 14 August 2025
- Published by Wolters Kluwer Health in Journal of Bone and Joint Surgery
- Vol. 107 (18) , 2102-2109
Impact of Childhood Obesity on Capital Femoral Epiphysis Morphology
Background: The precise pathological mechanisms through which obesity increases the risk of slipped capital femoral epiphysis (SCFE) remain unclear. We aimed to investigate the impact of childhood obesity on the morphology of the capital femoral epiphysis in children and adolescents without hip disorders. Methods: We performed a retrospective cross-sectional study that included 4,888 children and adolescents 7 to 19 years of age who underwent a hip or pelvic computed tomography (CT) scan, most for abdominal pain, between 2004 and 2022. Automated 3D-CT analysis assessed epiphyseal tubercle height, superior and anterior peripheral cupping, and epiphyseal tilt. Generalized additive models (GAMs) were used to examine the associations between epiphyseal morphology and age, stratified by obesity status, while adjusting for sex. Results: We observed distinct age-related trajectories of femoral-head morphology related to obesity. Subjects with obesity demonstrated a larger femoral-head diameter from ages 7 to 15 years (obese versus normal-weight differences: 0.3 to 2.3 mm), smaller epiphyseal tubercle height after 9 years of age (normal-weight versus obese differences: 0.4% to 1.3%), and a more posteriorly tilted epiphysis, as assessed by a smaller axial tilt angle (normal-weight versus obese differences: 4° to 1°) from ages 7 to 15 years. In subjects with obesity, the superior epiphyseal cupping was larger before 11 years of age (obese versus normal-weight differences: 1.6% to 0.4%), but after 13 years of age, the superior epiphyseal cupping was smaller (normal-weight versus obese differences: 0.2% to 2%). Conclusions: Children with obesity demonstrated a smaller epiphyseal tubercle height, greater posterior epiphyseal tilt, and reduced superior cupping compared with children of normal weight. These anatomical differences may contribute to the increased risk of SCFE in patients with obesity and offer potential imaging markers for earlier identification and risk-stratification. Level of Evidence: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
- 14 August 2025
- Published by Wolters Kluwer Health in Journal of Bone and Joint Surgery
- Vol. 107 (18) , 2102-2109
Impact of Childhood Obesity on Capital Femoral Epiphysis Morphology
Background: The precise pathological mechanisms through which obesity increases the risk of slipped capital femoral epiphysis (SCFE) remain unclear. We aimed to investigate the impact of childhood obesity on the morphology of the capital femoral epiphysis in children and adolescents without hip disorders. Methods: We performed a retrospective cross-sectional study that included 4,888 children and adolescents 7 to 19 years of age who underwent a hip or pelvic computed tomography (CT) scan, most for abdominal pain, between 2004 and 2022. Automated 3D-CT analysis assessed epiphyseal tubercle height, superior and anterior peripheral cupping, and epiphyseal tilt. Generalized additive models (GAMs) were used to examine the associations between epiphyseal morphology and age, stratified by obesity status, while adjusting for sex. Results: We observed distinct age-related trajectories of femoral-head morphology related to obesity. Subjects with obesity demonstrated a larger femoral-head diameter from ages 7 to 15 years (obese versus normal-weight differences: 0.3 to 2.3 mm), smaller epiphyseal tubercle height after 9 years of age (normal-weight versus obese differences: 0.4% to 1.3%), and a more posteriorly tilted epiphysis, as assessed by a smaller axial tilt angle (normal-weight versus obese differences: 4° to 1°) from ages 7 to 15 years. In subjects with obesity, the superior epiphyseal cupping was larger before 11 years of age (obese versus normal-weight differences: 1.6% to 0.4%), but after 13 years of age, the superior epiphyseal cupping was smaller (normal-weight versus obese differences: 0.2% to 2%). Conclusions: Children with obesity demonstrated a smaller epiphyseal tubercle height, greater posterior epiphyseal tilt, and reduced superior cupping compared with children of normal weight. These anatomical differences may contribute to the increased risk of SCFE in patients with obesity and offer potential imaging markers for earlier identification and risk-stratification. Level of Evidence: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.
- Open Access
- 4 March 2025
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 13 (3)
Effect of Initial Graft Tension on Knee Osteoarthritis Outcomes After ACL Reconstruction: A Randomized Controlled Clinical Trial With 15-Year Follow-up
Background: The graft tension applied during anterior cruciate ligament (ACL) graft fixation (subsequently referred to as initial graft tension) could potentiate posttraumatic osteoarthritis (PTOA) and influence other outcomes. Purpose/Hypothesis: The purpose of this study was to analyze the effects of initial graft tension on imaging and patient-reported outcomes related to PTOA 15 years postreconstruction surgery. Clinical and functional outcomes were also assessed. It was hypothesized that (1) the high-tension group would have improved imaging findings and outcomes compared with the low-tension group at 15-year follow-up and (2) the imaging findings and outcomes for the high-tension group would be equivalent to the sex-, race-, age-, and activity level matched control group. Study Design: Randomized controlled trial; Level of evidence, 1. Methods: Consented patients underwent ACL reconstruction (ACLR) with bone–patellar tendon–bone or a 4-strand hamstring tendon autograft. A matched uninjured control group was assembled for comparison. Two laxity-based tensioning procedures were randomized: (1) tension set to restore normal anteroposterior (AP) laxity at time of surgery relative to the contralateral uninjured knee (low-tension group) or (2) tension set to overconstrain AP laxity at time of surgery relative to the contralateral knee (high-tension group). Baseline outcome measures, radiographs, and magnetic resonance imaging (MRI) scans were collected prior to surgery and at 15 years postoperatively. Results: For medial joint space width, the differences between limbs across the 3 groups were not significant. Within the tension groups, the Whole Organ MRI Score was significantly higher in the surgical knee relative to the contralateral knee in both groups, while the Osteoarthritis Research Society International radiographic scores were higher in the surgical knee in the low-tension group only. A total of 43% of patients in both tension groups met the Knee injury and Osteoarthritis Outcome Score composite criteria for a symptomatic knee compared with controls (10%; P = .01). Most other outcomes, including AP laxity, International Knee Documentation Committee knee examination score, and single-leg hop test were not significantly different between the 3 groups. Conclusion: The results do not support the hypotheses that patients in the high-tension group would have better chondroprotection compared with the low-tension group and have equivalent outcomes with the matched controls. Overall, the results show that patients undergoing ACLR are more likely to develop PTOA and display inferior outcomes compared with the uninjured matched control group, regardless of graft tension.
- 4 March 2025
- Published by SAGE Publications in Orthopaedic Journal of Sports Medicine
- Vol. 13 (3)
Open AccessEffect of Initial Graft Tension on Knee Osteoarthritis Outcomes After ACL Reconstruction: A Randomized Controlled Clinical Trial With 15-Year Follow-up
Background: The graft tension applied during anterior cruciate ligament (ACL) graft fixation (subsequently referred to as initial graft tension) could potentiate posttraumatic osteoarthritis (PTOA) and influence other outcomes. Purpose/Hypothesis: The purpose of this study was to analyze the effects of initial graft tension on imaging and patient-reported outcomes related to PTOA 15 years postreconstruction surgery. Clinical and functional outcomes were also assessed. It was hypothesized that (1) the high-tension group would have improved imaging findings and outcomes compared with the low-tension group at 15-year follow-up and (2) the imaging findings and outcomes for the high-tension group would be equivalent to the sex-, race-, age-, and activity level matched control group. Study Design: Randomized controlled trial; Level of evidence, 1. Methods: Consented patients underwent ACL reconstruction (ACLR) with bone–patellar tendon–bone or a 4-strand hamstring tendon autograft. A matched uninjured control group was assembled for comparison. Two laxity-based tensioning procedures were randomized: (1) tension set to restore normal anteroposterior (AP) laxity at time of surgery relative to the contralateral uninjured knee (low-tension group) or (2) tension set to overconstrain AP laxity at time of surgery relative to the contralateral knee (high-tension group). Baseline outcome measures, radiographs, and magnetic resonance imaging (MRI) scans were collected prior to surgery and at 15 years postoperatively. Results: For medial joint space width, the differences between limbs across the 3 groups were not significant. Within the tension groups, the Whole Organ MRI Score was significantly higher in the surgical knee relative to the contralateral knee in both groups, while the Osteoarthritis Research Society International radiographic scores were higher in the surgical knee in the low-tension group only. A total of 43% of patients in both tension groups met the Knee injury and Osteoarthritis Outcome Score composite criteria for a symptomatic knee compared with controls (10%; P = .01). Most other outcomes, including AP laxity, International Knee Documentation Committee knee examination score, and single-leg hop test were not significantly different between the 3 groups. Conclusion: The results do not support the hypotheses that patients in the high-tension group would have better chondroprotection compared with the low-tension group and have equivalent outcomes with the matched controls. Overall, the results show that patients undergoing ACLR are more likely to develop PTOA and display inferior outcomes compared with the uninjured matched control group, regardless of graft tension.