Development and Validation of an Artificial Intelligence Model for Detecting Rib Fractures on Chest Radiographs.

Background: Chest radiography is the standard method for detecting rib fractures. Our study aims to develop an artificial intelligence (AI) model that, with only a relatively small amount of training data, can identify rib fractures on chest radiographs and accurately mark their precise locations, thereby achieving a diagnostic accuracy comparable to that of medical professionals. Methods: For this retrospective study, we developed an AI model using 540 chest radiographs (270 normal and 270 with rib fractures) labeled for use with Detectron2 which incorporates a faster region-based convolutional neural network (R-CNN) enhanced with a feature pyramid network (FPN). The model's ability to classify radiographs and detect rib fractures was assessed. Furthermore, we compared the model's performance to that of 12 physicians, including six board-certified anesthesiologists and six residents, through an observer performance test. Results: Regarding the radiographic classification performance of the AI model, the sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) were 0.87, 0.83, and 0.89, respectively. In terms of rib fracture detection performance, the sensitivity, false-positive rate, and free-response receiver operating characteristic (JAFROC) figure of merit (FOM) were 0.62, 0.3, and 0.76, respectively. The AI model showed no statistically significant difference in the observer performance test compared to 11 of 12 and 10 of 12 physicians, respectively. Conclusions: We developed an AI model trained on a limited dataset that demonstrated a rib fracture classification and detection performance comparable to that of an experienced physician.

Therapy-related myeloid neoplasm in early breast cancer patients treated with adjuvant chemotherapy.

BACKGROUND: Long-term complications are becoming more important as the survival rate of breast cancer improves. Treatment-related myeloid neoplasm is an important long-term complication in breast cancer survivors as it has a poor prognosis. OBJECTIVE: We evaluated the incidence and risk factors for the development of treatment-related acute myeloid leukaemia (AML)/myelodysplastic syndrome (MDS) in patients treated with early breast cancer. METHODS: We accessed the national Korean database to identify 153,565 patients diagnosed with breast cancer between January 2007 and October 2016 who underwent surgery for breast cancer. We estimated the cumulative incidence of AML/MDS and analysed the risk factors for developing AML/MDS. RESULTS: Of 153,575 patients, 79,321 received anthracycline-based adjuvant therapy, 14,317 received adjuvant therapy without anthracyclines and 46,657 did not receive adjuvant chemotherapy. Overall, 120 developed AML (105 in the anthracycline group, 9 in the non-anthracycline group and 6 in the control group), and 128 developed MDS (96, 9 and 23 in each group). The 10-year cumulative incidence of AML/MDS was the highest in the anthracycline group (0.221% and 0.199%), followed by the non-anthracycline group (0.122% and 0.163%) and the control group (0.024% and 0.089%). The risk of developing AML/MDS was significantly higher in patients treated with anthracyclines (hazard ratio [HR] 9.531; p < 0.0001 for AML and HR 2.559; p < 0.0001 for MDS) compared to patients in the control group. CONCLUSION: This study found that anthracycline-based adjuvant therapy significantly increased the risk of AML/MDS in Korean breast cancer patients, with the risk persisting for at least 10 years. While the cumulative incidence was low, the long-term risks of AML/MDS should be taken into account considering the poor outcomes associated with these neoplasms.

Comparisons of Prehospital Delay and Related Factors Between Acute Ischemic Stroke and Acute Myocardial Infarction.

Background Prehospital delay is an important contributor to poor outcomes in both acute ischemic stroke (AIS) and acute myocardial infarction (AMI). We aimed to compare the prehospital delay and related factors between AIS and AMI. Methods and Results We identified patients with AIS and AMI who were admitted to the 11 Korean Regional Cardiocerebrovascular Centers via the emergency room between July 2016 and December 2018. Delayed arrival was defined as a prehospital delay of >3 hours, and the generalized linear mixed-effects model was applied to explore the effects of potential predictors on delayed arrival. This study included 17 895 and 8322 patients with AIS and AMI, respectively. The median value of prehospital delay was 6.05 hours in AIS and 3.00 hours in AMI. The use of emergency medical services was the key determinant of delayed arrival in both groups. Previous history, 1-person household, weekday presentation, and interhospital transfer had higher odds of delayed arrival in both groups. Age and sex had no or minimal effects on delayed arrival in AIS; however, age and female sex were associated with higher odds of delayed arrival in AMI. More severe symptoms had lower odds of delayed arrival in AIS, whereas no significant effect was observed in AMI. Off-hour presentation had higher and prehospital awareness had lower odds of delayed arrival; however, the magnitude of their effects differed quantitatively between AIS and AMI. Conclusions The effects of some nonmodifiable and modifiable factors on prehospital delay differed between AIS and AMI. A differentiated strategy might be required to reduce prehospital delay.

Risk of reoperation and infection after percutaneous endoscopic lumbar discectomy and open lumbar discectomy : a nationwide population-based study.

AIMS: Open discectomy (OD) is the standard operation for lumbar disc herniation (LDH). Percutaneous endoscopic lumbar discectomy (PELD), however, has shown similar outcomes to OD and there is increasing interest in this procedure. However despite improved surgical techniques and instrumentation, reoperation and infection rates continue and are reported to be between 6% and 24% and 0.7% and 16%, respectively. The objective of this study was to compare the rate of reoperation and infection within six months of patients being treated for LDH either by OD or PELD. METHODS: In this retrospective, nationwide cohort study, the Korean National Health Insurance database from 1 January 2007 to 31 December 2018 was reviewed. Data were extracted for patients who underwent OD or PELD for LDH without a history of having undergone either procedure during the preceding year. Individual patients were followed for six months through their encrypted unique resident registration number. The primary endpoints were rates of reoperation and infection during the follow-up period. Other risk factors for reoperation and infection were also evalulated. RESULTS: Out of 549,531 patients, 522,640 had undergone OD (95.11%) and 26,891 patients had undergone PELD (4.89%). Reoperation rates within six months were 2.28% in the OD group, and 5.38% in the PELD group. Infection rates were 1.18% in OD group and 0.83% in PELD group. The risk of reoperation was lower for patients with OD than for patients with PELD (adjusted hazard ratio (HR) 0.38). The risk of infection was higher for patients with OD than for patients undergoing PELD (HR, 1.325). CONCLUSION: Compared with the OD group, the PELD group showed higher reoperation rates and lower infection rates. Cite this article: Bone Joint J 2021;103-B(8):1392-1399.