Development and validation of an artificial intelligence model for the classification of hip fractures using the AO-OTA framework.

BACKGROUND AND PURPOSE: Artificial intelligence (AI) has the potential to aid in the accurate diagnosis of hip fractures and reduce the workload of clinicians. We primarily aimed to develop and validate a convolutional neural network (CNN) for the automated classification of hip fractures based on the 2018 AO-OTA classification system. The secondary aim was to incorporate the model's assessment of additional radiographic findings that often accompany such injuries. METHODS: 6,361 plain radiographs of the hip taken between 2002 and 2016 at Danderyd University Hospital were used to train the CNN. A separate set of 343 radiographs representing 324 unique patients was used to test the performance of the network. Performance was evaluated using area under the curve (AUC), sensitivity, specificity, and Youden's index. RESULTS: The CNN demonstrated high performance in identifying and classifying hip fracture, with AUCs ranging from 0.76 to 0.99 for different fracture categories. The AUC for hip fractures ranged from 0.86 to 0.99, for distal femur fractures from 0.76 to 0.99, and for pelvic fractures from 0.91 to 0.94. For 29 of 39 fracture categories, the AUC was ≥ 0.95. CONCLUSION: We found that AI has the potential for accurate and automated classification of hip fractures based on the AO-OTA classification system. Further training and modification of the CNN may enable its use in clinical settings.

Artificial intelligence can be used in the identification and classification of shoulder osteoarthritis and avascular necrosis on plain radiographs: a training study of 7,139 radiograph sets.

BACKGROUND AND PURPOSE: Knowledge concerning the use AI models for the classification of glenohumeral osteoarthritis (GHOA) and avascular necrosis (AVN) of the humeral head is lacking. We aimed to analyze how a deep learning (DL) model trained to identify and grade GHOA on plain radiographs performs. Our secondary aim was to train a DL model to identify and grade AVN on plain radiographs. PATIENTS AND METHODS: A modified ResNet-type network was trained on a dataset of radiographic shoulder examinations from a large tertiary hospital. A total of 7,139 radiographs were included. The dataset included various projections of the shoulder, and the network was trained using stochastic gradient descent. Performance evaluation metrics, area under the receiver operating characteristic curve (AUC), sensitivity, and specificity were used to assess the network's performance for each outcome. RESULTS: The network demonstrated AUC values ranging from 0.73 to 0.93 for GHOA classification and > 0.90 for all AVN classification classes. The network exhibited lower AUC for mild cases compared with definitive cases of GHOA. When none and mild grades were combined, the AUC increased, suggesting difficulties in distinguishing between these 2 grades. CONCLUSION: We found that a DL model can be trained to identify and grade GHOA on plain radiographs. Furthermore, we show that a DL model can identify and grade AVN on plain radiographs. The network performed well, particularly for definitive cases of GHOA and any level of AVN. However, challenges remain in distinguishing between none and mild GHOA grades.

Plasma H3Cit-DNA Discriminates Between Cancer and Inflammation in a Cohort of Patients with Unspecific Cancer Symptoms.

Cancer detection is challenging, especially in patients with unspecific cancer symptoms. Biomarkers could identify patients at high risk of cancer. Prior studies indicate that neutrophil extracellular traps (NETs) are associated with cancer, but also with autoimmune and infectious diseases. The objective of this prospective study was to investigate markers associated with NET formation (nucleosomal citrullinated histone 3 [H3Cit-DNA], cell free DNA [cfDNA] and neutrophil elastase [NE]), and c-reactive protein (CRP) in patients with unspecific cancer symptoms, such as fatigue, weight loss or radiological sign of malignancy without an apparent primary tumor, referred to the Diagnostic Center at Danderyd Hospital in Sweden. Blood samples were drawn on admission, before cancer diagnosis. Out of 475 patients, 160 (34%) were diagnosed with cancer, 56 (12%) with autoimmune disease, 32 (7%) with infectious disease, 71 (15%) with other diseases and 156 (33%) received no diagnosis. H3Cit-DNA, cfDNA, NE and CRP were significantly higher in patients with cancer compared to patients without cancer (p < 0.0001, p < 0.0001, p = 0.004, and p = 0.0002 respectively). H3Cit-DNA, but not cfDNA, NE or CRP, was significantly elevated in patients with cancer compared to patients with autoimmune disease (p = 0.0001). H3Cit-DNA, cfDNA, NE or CRP did not differ between cancer and infectious disease. In conclusion, H3Cit-DNA is elevated in patients diagnosed with cancer compared to non-cancer patients with the same symptomatology. Further studies should evaluate if H3Cit-DNA could aid in selecting patients that would benefit the most from a rapid cancer diagnostic work-up.

Modulation of FGF pathway signaling and vascular differentiation using designed oligomeric assemblies.

Many growth factors and cytokines signal by binding to the extracellular domains of their receptors and driving association and transphosphorylation of the receptor intracellular tyrosine kinase domains, initiating downstream signaling cascades. To enable systematic exploration of how receptor valency and geometry affect signaling outcomes, we designed cyclic homo-oligomers with up to 8 subunits using repeat protein building blocks that can be modularly extended. By incorporating a de novo-designed fibroblast growth factor receptor (FGFR)-binding module into these scaffolds, we generated a series of synthetic signaling ligands that exhibit potent valency- and geometry-dependent Ca2+ release and mitogen-activated protein kinase (MAPK) pathway activation. The high specificity of the designed agonists reveals distinct roles for two FGFR splice variants in driving arterial endothelium and perivascular cell fates during early vascular development. Our designed modular assemblies should be broadly useful for unraveling the complexities of signaling in key developmental transitions and for developing future therapeutic applications.

ChatGPT-4 generates orthopedic discharge documents faster than humans maintaining comparable quality: a pilot study of 6 cases.

BACKGROUND AND PURPOSE: Large language models like ChatGPT-4 have emerged. They hold the potential to reduce the administrative burden by generating everyday clinical documents, thus allowing the physician to spend more time with the patient. We aimed to assess both the quality and efficiency of discharge documents generated by ChatGPT-4 in comparison with those produced by physicians. PATIENTS AND METHODS: To emulate real-world situations, the health records of 6 fictional orthopedic cases were created. Discharge documents for each case were generated by a junior attending orthopedic surgeon and an advanced orthopedic resident. ChatGPT-4 was then prompted to generate the discharge documents using the same health record information. The quality assessment was performed by an expert panel (n = 15) blinded to the source of the documents. As secondary outcome, the time required to generate the documents was compared, logging the duration of the creation of the discharge documents by the physician and by ChatGPT-4. RESULTS: Overall, both ChatGPT-4 and physician-generated notes were comparable in quality. Notably, ChatGPT-4 generated discharge documents 10 times faster than the traditional method. 4 events of hallucinations were found in the ChatGPT-4-generated content, compared with 6 events in the human/physician produced notes. CONCLUSION: ChatGPT-4 creates orthopedic discharge notes faster than physicians, with comparable quality. This shows it has great potential for making these documents more efficient in orthopedic care. ChatGPT-4 has the potential to significantly reduce the administrative burden on healthcare professionals.

Comparison of Comorbidity Models Within a Population-Based Cohort of Older Adults With Non-Hodgkin Lymphoma.

PURPOSE: Compare the association of individual comorbidities, comorbidity indices, and survival in older adults with non-Hodgkin lymphoma (NHL), including in specific NHL subtypes. METHODS: Data source was SEER-Medicare, a population-based registry of adults age 65 years and older with cancer. We included all incident cases of NHL diagnosed during 2008-2017 who met study inclusion criteria. Comorbidities were classified using the three-factor risk estimate scale (TRES), Charlson comorbidity index (CCI), and National Cancer Institute (NCI) comorbidity index categories and weights. Overall survival (OS) and lymphoma-specific survival, with death from other causes treated as a competing risk, were estimated using the Kaplan-Meier method from time of diagnosis. Multivariable Cox models were constructed, and Harrel C-statistics were used to compare comorbidity models. A two-sided P value of <.05 was considered significant. RESULTS: A total of 40,486 patients with newly diagnosed NHL were included. Patients with aggressive NHL had higher rates of baseline comorbidity. Despite differences in baseline comorbidity between NHL subtypes, cardiovascular, pulmonary, diabetes, and renal comorbidities were frequent and consistently associated with OS in most NHL subtypes. These categories were used to construct a candidate comorbidity score, the non-Hodgkin lymphoma 5 (NHL-5). Comparing three validated comorbidity scores, TRES, CCI, NCI, and the novel NHL-5 score, we found similar associations with OS and lymphoma-specific survival, which was confirmed in sensitivity analyses by NHL subtypes. CONCLUSION: The optimal measure of comorbidity in NHL is unknown. Here, we demonstrate that the three-category TRES and five-category NHL-5 scores perform as well as the 14-16 category CCI and NCI scores in terms of association with OS and lymphoma-specific survival. These simple scores could be more easily used in clinical practice without prognostic loss.

Effect of a national infection control programme in Sweden on prosthetic joint infection incidence following primary total hip arthroplasty: a cohort study.

OBJECTIVES: Prosthetic joint infection (PJI) is a serious complication following total hip arthroplasty (THA) entailing increased mortality, decreased quality of life and high healthcare costs.The primary aim was to investigate whether the national project: Prosthesis Related Infections Shall be Stopped (PRISS) reduced PJI incidence after primary THA; the secondary aim was to evaluate other possible benefits of PRISS, such as shorter time to diagnosis. DESIGN: Cohort study. SETTING: In 2009, a nationwide, multidisciplinary infection control programme was launched in Sweden, PRISS, which aimed to reduce the PJI burden by 50%. PARTICIPANTS: We obtained data on patients undergoing primary THA from the Swedish Arthroplasty Registry 2012-2014, (n=45 723 patients, 49 946 THAs). Using personal identity numbers, this cohort was matched with the Swedish Prescribed Drug Registry. Medical records of patients with ≥4 weeks' antibiotic consumption were reviewed to verify PJI diagnosis (n=2240, 2569 THAs). RESULTS: The cumulative incidence of PJI following the PRISS Project was 1.2% (95% CI 1.1% to 1.3%) as compared with 0.9% (95% CI 0.8% to 1.0%) before. Cox regression models for the PJI incidence post-PRISS indicates there was no statistical significance difference versus pre-PRISS (HR 1.1 (95% CI 0.9 to 1.3)). There was similar time to PJI diagnosis after the PRISS Project 24 vs 23 days (p=0.5). CONCLUSIONS: Despite the comprehensive nationwide PRISS Project, Swedish PJI incidence was higher after the project and time to diagnosis remained unchanged. Factors contributing to PJI, such as increasing obesity, higher American Society of Anesthesiology class and more fractures as indications, explain the PJI increase among primary THA patients.

Optimization of mHealth behavioral interventions for patients with chronic lymphocytic leukemia: the HEALTH4CLL study.

PURPOSE: This pilot study of a diet and physical activity intervention (HEALTH4CLL) was conducted to reduce fatigue and improve physical function (PF) in patients with chronic lymphocytic leukemia (CLL). METHODS: The HEALTH4CLL study used a randomized factorial design based on the multiphase optimization strategy (MOST). Patients received diet, exercise, and body weight management instructional materials plus a Fitbit and were randomized to undergo one of 16 combinations of 4 evidence-based mHealth intervention strategies over 16 weeks. Patients' fatigue, PF, health-related quality of life, behavior changes, and program satisfaction and retention were assessed. Paired t-tests were used to examine changes in outcomes from baseline to follow-up among patients. Factorial analysis of variance examined effective intervention components and their combinations regarding improvement in fatigue and PF scores. RESULTS: Among 31 patients, we observed significant improvements in fatigue (+ 11.8; t = 4.08, p = 0.001) and PF (+ 2.6; t = 2.75, p = 0.01) scores. The combination of resistance and aerobic exercise with daily self-monitoring was associated with improved fatigue scores (ß = 3.857, SE = 1.617, p = 0.027). Analysis of the individual components of the MOST design demonstrated greater improvement in the PF score with resistance plus aerobic exercise than with aerobic exercise alone (ß = 2.257, SE = 1.071, p = 0.048). CONCLUSIONS: Combined aerobic and resistance exercise and daily self-monitoring improved PF and reduced fatigue in patients with CLL. IMPLICATIONS FOR CANCER SURVIVORS: This pilot study supported the feasibility of a low-touch mHealth intervention for survivors of CLL and provided preliminary evidence that exercising, particularly resistance exercise, can improve their symptoms and quality of life.

Timing of Surgery for Proximal Humeral Fracture Treated with Shoulder Hemiarthroplasty, Best Results with Surgery Within 2 Weeks.

Background: Preoperative delay may affect the outcome of proximal humerus fractures treated with shoulder hemiarthroplasty. There is currently no consensus for the recommended preoperative time interval. The aim was to examine how the time to surgery with shoulder hemiarthroplasty after a proximal humerus fracture affected the patient-reported outcome. Methods: 380 patients with proximal humerus fractures treated with shoulder hemiarthroplasty recorded from the Swedish Shoulder Arthroplasty Registry were included. Three self-reporting outcome instruments were used at follow-up after 1-5 years: a shoulder-specific score, the Western Ontario Osteoarthritis of the Shoulder index (WOOS), the EuroQol-5 Dimension index (EQ-5D), and subjective patient satisfaction assessment. Results: The preoperative delay had a negative impact on the WOOS, EQ-5D, and patient satisfaction level (p < 0.01). The best result, measured with WOOS at a minimum 1-year follow-up, was found when surgery was performed 6-10 days after the reported date of fracture. WOOS% 8-14 days was 69.4% (± 24.2). A delay of more than 10 days was shown to be correlated with poorer outcomes. WOOS% 15-60 days was 55.8% (± 25.0) and continued to decrease. Conclusion: The current recommendation in Sweden to perform shoulder hemiarthroplasty within 2 weeks after sustaining a proximal humerus fracture is considered valid.

Systematic review of machine-learning models in orthopaedic trauma.

Aims: Machine-learning (ML) prediction models in orthopaedic trauma hold great promise in assisting clinicians in various tasks, such as personalized risk stratification. However, an overview of current applications and critical appraisal to peer-reviewed guidelines is lacking. The objectives of this study are to 1) provide an overview of current ML prediction models in orthopaedic trauma; 2) evaluate the completeness of reporting following the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement; and 3) assess the risk of bias following the Prediction model Risk Of Bias Assessment Tool (PROBAST) tool. Methods: A systematic search screening 3,252 studies identified 45 ML-based prediction models in orthopaedic trauma up to January 2023. The TRIPOD statement assessed transparent reporting and the PROBAST tool the risk of bias. Results: A total of 40 studies reported on training and internal validation; four studies performed both development and external validation, and one study performed only external validation. The most commonly reported outcomes were mortality (33%, 15/45) and length of hospital stay (9%, 4/45), and the majority of prediction models were developed in the hip fracture population (60%, 27/45). The overall median completeness for the TRIPOD statement was 62% (interquartile range 30 to 81%). The overall risk of bias in the PROBAST tool was low in 24% (11/45), high in 69% (31/45), and unclear in 7% (3/45) of the studies. High risk of bias was mainly due to analysis domain concerns including small datasets with low number of outcomes, complete-case analysis in case of missing data, and no reporting of performance measures. Conclusion: The results of this study showed that despite a myriad of potential clinically useful applications, a substantial part of ML studies in orthopaedic trauma lack transparent reporting, and are at high risk of bias. These problems must be resolved by following established guidelines to instil confidence in ML models among patients and clinicians. Otherwise, there will remain a sizeable gap between the development of ML prediction models and their clinical application in our day-to-day orthopaedic trauma practice.

Functional outcomes are restored a decade after a distal radius fracture: a prospective long-term follow-up study.

We performed an 11-13-year prospective follow-up of patients after a distal radial fracture (DRF) to investigate the association between fracture malunion, radiocarpal osteoarthritis and clinical outcome. In total, 292 patients responded to patient-reported outcome measures; of them, 242 underwent clinical examination. Clinical outcomes improved with time. A decade after fracture, median Quick Disabilities of the Arm, Shoulder and Hand (QuickDASH) score was 5, EuroQol Five-dimensions score was 1.0, and range of motion and grip strength were 96% of the contralateral side. Neither osteoarthritis (6%) nor pseudoarthrosis of the ulnar styloid (30%) affected the outcomes. Dorsal tilt, radial inclination, ulnar variance and intra-articular extension did not affect long-term clinical outcomes or the risk of osteoarthritis. Recovery after a DRF is an ongoing process that lasts years. A decade after the injury event, range of motion, grip strength and QuickDASH were recovered to population normal, regardless of radiological outcomes.Level of evidence: II.

Safety and efficacy of ibrutinib in combination with rituximab and lenalidomide in previously untreated follicular and marginal zone lymphoma: An open label, phase 2 study.

BACKGROUND: Follicular lymphoma (FL) and marginal zone lymphoma (MZL) are indolent non-Hodgkin lymphomas (iNHL). Median survival for iNHL is approximately 20 years. Because standard treatments are not curative, patients often receive multiple lines of therapy with associated toxicity-rationally designed, combination therapies with curative potential are needed. The immunomodulatory drug lenalidomide was evaluated in combination with rituximab for the frontline treatment of FL in the phase 3 RELEVANCE study. Ibrutinib, an oral Bruton tyrosine kinase inhibitor, is active in NHL and was evaluated in combination with lenalidomide, rituximab, and ibrutinib (IRR) in a phase 1 study. METHODS: The authors conducted an open-label, phase 2 clinical trial of IRR for previously untreated FL and MZL. The primary end point was progression-free survival (PFS) at 24 months. RESULTS: This study included 48 participants with previously untreated FL grade 1-3a (N = 38), or MZL (N = 10). Participants received 12, 28-day cycles of lenalidomide (15 mg, days 1-21 cycle 1; 20 mg, cycles 2-12), rituximab (375 mg/m2 weekly in cycle 1; day 1 cycles 2-12), and ibrutinib 560 mg daily. With a median follow-up of 65.3 months, the estimated PFS at 24 months was 78.8% (95% confidence interval [CI], 68.0%-91.4%) and 60-month PFS was 59.7% (95% CI, 46.6%-76.4%). One death occurred unrelated to disease progression. Grade 3-4 adverse events were observed in 64.6%, including 50% with grade 3-4 rash. CONCLUSIONS: IRR is highly active as frontline therapy for FL and MZL. Compared to historical results with lenalidomide and rituximab, PFS is similar with higher grade 3-4 toxicity, particularly rash. The study was registered with ClinicalTrials.gov (NCT02532257).

Influence of treatment intensity and medical comorbidities in older adults with peripheral T cell lymphoma.

We conducted a population-based study of patients >65 years, diagnosed 2008-2017, with peripheral T-cell lymphoma (PTCL) using SEER-Medicare. Associations between PTCL subtype, treatment regimen, comorbidity, and mortality were assessed using the Kaplan-Meier method and multivariable Cox regression. Amongst the 2,546 patients, the median age was 77 years (interquartile range, 71-83). 5-year overall survival (OS) ranged from 22.2% to 37.3% depending on PTCL subtype. The most common frontline regimen was cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP). 5-year OS rate was 47.0% for patients treated with etoposide + CHOP (N = 67; CHOEP), 33.7% for those treated with CHOP (N = 732), and 23.8% for patients treated with non-anthracycline-containing regimens (N = 105; p < 0.001). In patients without comorbidities, CHOEP remained independently associated with improved OS (HR 0.52, 95% CI,0.30-0.91). Median OS was 1.2 years from initiation of second-line therapy (N = 228) independent of treatment regimen. Frontline but not second-line treatment regimen is associated with OS in older patients with PTCL.

A single-cell atlas of CD19 chimeric antigen receptor T cells.

Li et al. present a resource of single-cell RNA sequencing (scRNA-seq) data from the infusion products of relapsed or refractory large B cell lymphoma (rrLBCL) patients treated with standard-of-care axicabtagene ciloleucel and identify features that are significantly different between products from responders and non-responders at 3-month followup by PET/CT, an important landmark for long-term outcomes.

Increased mortality after total hip prosthetic joint infection is mainly caused by the comorbidities rather than the infection itself.

BACKGROUND AND PURPOSE: Periprosthetic joint infection (PJI) is a feared complication of arthroplasty surgery. There is controversy as to whether PJI also correlates with increased mortality. Our aim was to investigate in a nationwide cohort if PJI is an independent risk factor for dying. PATIENTS AND METHODS: We performed a retrospective cohort study based on data from the Swedish Hip Arthroplasty Register (SHAR). All patients with a revision THA performed between 1998 and 2017 were included. The outcome is mortality; exposure is PJI according to SHAR. The control group was study participants who underwent aseptic revision. Confounders were age, sex, diagnosis, and comorbidity according to the Elixhauser index. The outcome was analyzed with a Cox proportional hazards model. RESULTS: 4,943 PJI revisions and 12,529 non-infected revisions were included in the analysis. The median follow-up time was 4.1 years. In the PJI group, 1,972 patients died and in the control group, 4,512. The incidence rate ratio was 1.19 (95% confidence interval [CI] 1.13-1.25), the crude hazard ratio (HR) 1.19 (CI 1.13-1.25), and the adjusted HR 1.05 (CI 0.99-1.12) for the exposed versus the unexposed group. The strongest confounder was comorbidity. CONCLUSION: The increased mortality risk after revision due to PJI is mainly caused by the comorbidity of the patient, rather than by the infection itself.

Deep learning classification of shoulder fractures on plain radiographs of the humerus, scapula and clavicle.

In this study, we present a deep learning model for fracture classification on shoulder radiographs using a convolutional neural network (CNN). The primary aim was to evaluate the classification performance of the CNN for proximal humeral fractures (PHF) based on the AO/OTA classification system. Secondary objectives included evaluating the model's performance for diaphyseal humerus, clavicle, and scapula fractures. The training dataset consisted of 6,172 examinations, including 2-7 radiographs per examination. The overall area under the curve (AUC) for fracture classification was 0.89, indicating good performance. For PHF classification, 12 out of 16 classes achieved an AUC of 0.90 or greater. Additionally, the CNN model had excellent overall AUC for diaphyseal humerus fractures (0.97), clavicle fractures (0.96), and good AUC for scapula fractures (0.87). Despite the limitations of the study, such as the reliance on ground truth labels provided by students with limited radiographic assessment experience, our findings are in concordance with previous studies, further consolidating CNN as potent fracture classifiers in plain radiographs. The inclusion of multiple radiographs with different views from each examination, as well as the generally unselected nature of the sample, contributed to the overall generalizability of the study. This is the fifth study published by our group on AI in orthopaedic radiographs, which has consistently shown promising results. The next challenge for the orthopaedic research community will be to transfer these results from the research setting into clinical practice. External validation of the CNN model should be conducted in the future before it is considered for use in a clinical setting.

Timing of adverse events in patients undergoing acute and elective hip arthroplasty surgery: a multicentre cohort study using the Global Trigger Tool.

OBJECTIVE: To explore timing in relation to all types of adverse events (AEs), severity and preventability for patients undergoing acute and elective hip arthroplasty. DESIGN: A multicentre cohort study using retrospective record review with Global Trigger Tool methodology in combination with data from several registers. SETTING: 24 hospitals in 4 major regions of Sweden. PARTICIPANTS: Patients ≥18 years, undergoing acute or elective total or hemiarthroplasty of the hip, were eligible for inclusion. Reviews of weighted samples of 1998 randomly selected patient records were carried out using Global Trigger Tool methodology. The patients were followed for readmissions up to 90 days postoperatively throughout the whole country. RESULTS: The cohort consisted of 667 acute and 1331 elective patients. Most AEs occurred perioperatively and postoperatively (n=2093, 99.1%) and after discharge (n=1142, 54.1%). The median time from the day of surgery to the occurrence of AE was 8 days. The median days for different AE types ranged from 0 to 24.5 for acute and 0 to 71 for elective patients and peaked during different time periods. 40.2% of the AEs, both major and minor, occurred within postoperative days 0-5 and 86.9% of the AEs occurred within 30 days. Most of the AEs were deemed to be of major severity (n=1370, 65.5%) or preventable (n=1591, 76%). CONCLUSIONS: A wide variability was found regarding the timing of different AEs with the majority occurring within 30 days. The timing and preventability varied regarding the severity. Most of the AEs were deemed to be preventable and/or of major severity. To increase patient safety for patients undergoing hip arthroplasty surgery, a better understanding of the multifaceted nature of the timing of AEs in relation to the occurrence of differing AEs is needed.