Search

Emotion understanding is an essential but highly challenging component of artificial general intelligence. The absence of extensively annotated datasets has significantly impeded advancements in this field. We present EmotionCLIP, the first pre-training paradigm to extract visual emotion representations from verbal and nonverbal communication using only uncurated data. Compared to numerical labels or descriptions used in previous methods, communication naturally contains emotion information. Furthermore, acquiring emotion representations from communication is more congruent with the human learning process. We guide EmotionCLIP to attend to nonverbal emotion cues through subject-aware context encoding and verbal emotion cues using sentiment-guided contrastive learning. Extensive experiments validate the effectiveness and transferability of EmotionCLIP. Using merely linear-probe evaluation protocol, EmotionCLIP outperforms the state-of-the-art supervised visual emotion recognition methods and rivals many multimodal approaches across various benchmarks. We anticipate that the advent of EmotionCLIP will address the prevailing issue of data scarcity in emotion understanding, thereby fostering progress in related domains. The code and pre-trained models are available at https://github.com/Xeaver/EmotionCLIP., Comment: CVPR 2023

Additional file 3. Supplementary file 3. Cut-off values of the GAP score.

Additional file 2. Supplementary file 2. Ideal values of different parameters in Roussouly classification.

Additional file 1. Supplementary file 1. The definitions of radiographic parameters.

Wenzhi Sun, Shibao Lu, Chao Kong, Zhongen Li, Peng Wang, Sitao Zhang Department of Orthopaedics, Capital Medical University Xuanwu Hospital, Beijing 100053, People’s Republic of ChinaCorrespondence: Shibao Lu Email xuanwuspine@sina.comBackground and Aim: Frailty is an independent predictor of mortality and adverse events (AEs) in patients undergoing surgery. This study aimed to quantify the ability of Modified Frailty Index (mFI) to predict AEs in older patients undergoing elective posterior thoracolumbar fusion surgery.Methods: We retrospectively reviewed the results of 426 patients with the following diagnoses and follow-up evaluations of at least 12 months duration: lumbar disc herniation, 125; degenerative spondylolisthesis, 81; lumbar spinal canal stenosis, 187; and adult spinal deformities, 33. The cases were divided into two groups. The long spinal fusion (LSF) group was defined as ≥ 3 spinal levels with segmental pedicle-screw fixation. Short spinal fusion (SSF) were defined with at most two levels. The mFI used in the present study is an 11-variable assessment. The association of frailty with AEs was determined after adjusting for known and suspected confounders.Results: Frailty was presented in 66 patients (15.5%) within the total population (LSF, 21.9% and SSF, 11.8%). Rates of AEs assessed in the study increased stepwise with an increase in the mFI for the two groups. The severity of frailty was an independent predictor of any, major, and minor complications in the LSF group and any, minor complication in the SSF group (P< 0.05). A comparison of post-operative clinical outcomes showed that the ODI and SF-36 scores deteriorated as the mFI increased.Conclusion: Frailty was shown to be an independent predictor of AEs in older patients undergoing elective posterior thoracolumbar fusion surgery, especially for patients undergoing LSF.Keywords: frailty, degenerative spine disease, spinal surgery, elderly, clinical evaluation, complications, risk stratification