Scientific discovery in the age of artificial intelligence.

Artificial intelligence (AI) is being increasingly integrated into scientific discovery to augment and accelerate research, helping scientists to generate hypotheses, design experiments, collect and interpret large datasets, and gain insights that might not have been possible using traditional scientific methods alone. Here we examine breakthroughs over the past decade that include self-supervised learning, which allows models to be trained on vast amounts of unlabelled data, and geometric deep learning, which leverages knowledge about the structure of scientific data to enhance model accuracy and efficiency. Generative AI methods can create designs, such as small-molecule drugs and proteins, by analysing diverse data modalities, including images and sequences. We discuss how these methods can help scientists throughout the scientific process and the central issues that remain despite such advances. Both developers and users of AI toolsneed a better understanding of when such approaches need improvement, and challenges posed by poor data quality and stewardship remain. These issues cut across scientific disciplines and require developing foundational algorithmic approaches that can contribute to scientific understanding or acquire it autonomously, making them critical areas of focus for AI innovation.

Phillyrin ameliorates influenza a virus-induced pulmonary inflammation by antagonizing CXCR2 and inhibiting NLRP3 inflammasome activation.

Influenza is an acute viral respiratory illness with high morbidity rates worldwide. Excessive pulmonary inflammation is the main characteristic of lethal influenza A virus (IAV) infections. Therapeutic options for managing influenza are limited to vaccines and some antiviral medications. Phillyrin is one of the major bioactive components of the Chinese herbal medicine Forsythia suspensa, which has the functions of sterilization, heat clearing and detoxification. In this work, the effect and mechanism of phillyrin on H1N1 influenza (PR8)-induced pneumonia were investigated. We reported that phillyrin (15 mg/kg) treatment after viral challenge significantly improved the weight loss, ameliorated pulmonary inflammation and inhibited the accumulation of multiple cytokines and chemokines in bronchoalveolar lavage fluid on 7 days post infection (dpi). In vitro, phillyrin suppressed influenza viral replication (Matrixprotein and nucleoprotein messenger RNA level) and reduced influenza virus-induced cytopathic effect (CPE). Furthermore,chemokine receptor CXCR2 was confirmed to be markedly inhibited by phillyrin. Surface plasmon resonance results reveal that phillyrin exhibits binding affinity to CXCR2, having a binding affinity constant (KD) value of 1.858e-5 M, suggesting that CXCR2 is a potential therapeutic target for phillyrin. Moreover, phillyrin inhibited the mRNA and protein expression levels of Caspase1, ASC and NLRP3 in the lungs of mice with H1N1-induced pneumonia.This study reveals that phillyrin ameliorates IAV-induced pulmonary inflammation by antagonizing CXCR2 and inhibiting NLRP3 inflammasome activation partly.

Sparstolonin B inhibits pancreatic adenocarcinoma through the NF-κB signaling pathway.

Pancreatic adenocarcinoma is a highly lethal malignant gastrointestinal tumor. Sparstolonin B is an isocoumarin whose anticancer activity has recently received increasing attention. This study aimed to investigate Sparstolonin B's potential antitumor effect on pancreatic adenocarcinoma. The effect of Sparstolonin B on pancreatic cancer target genes and molecular mechanism was predicted via network pharmacology; Sparstolonin B significantly decreased Panc-1 and SW1990 cell viability and effectively suppressed the proliferation, migration, and invasion of pancreatic cancer cells as shown by CCK-8, colony formation, and Transwell assays. Flow cytometry showed that it induced cell cycle arrest and apoptosis. Sparstolonin B also upregulated Bax levels but decreased those of MMP2 and Bcl-2, downregulated IκBα expression, and upregulated p65 and IκBα phosphorylation; however, it had no effect on total NF-κB p65 levels. The NF-κB pathway inhibitor QNZ reversed these effects. The treatment group (26 µmol/L) had reduced graft volume and weight and fewer Ki-67-positive cells than the control group. Therefore, Sparstolonin B can inhibit the growth and induce the apoptosis of pancreatic cancer cells via the NF-κB signaling pathway and may be a potential novel drug for pancreatic cancer treatment.

Different effects of continuous and pulsed Benzo[a]pyrene exposure on metabolism and antioxidant defense of large yellow croaker: Depend on exposure duration.

This study aims to compare differential effects of continuous and pulsed BaP exposures on metabolism and antioxidant defense in the liver of large yellow croaker. Fish were subjected to BaP for 4 days and 36 days in three exposure regimes with the same time-averaged concentration of BaP: 4 µg/L BaP continuously, 8 µg/L BaP for 24 h every other day or 16 µg/L BaP for 24 h every 4 days. Our results showed that compared to pulsed BaP exposures, continuous BaP exposure reduced BaP metabolism (CYP1A, CYP3A and AHR transcriptional expressions, GSH content, GSH/GSSG ratio, EROD and GST activities) and antioxidant defense (T-SOD activity) on day 4, resulting to the increases in MDA and PC contents, indicating that continuous BaP exposure induced more severe oxidative damage during the early stage of exposure. But continuous BaP exposure reduced MDA and PC contents by improving BaP metabolism and antioxidant defense during the late stage of exposure. CYP1B transcriptional expression and CAT activity were unsuitable biomarkers of both continuous and pulsed BaP exposures. In conclusion, our results demonstrated differential effects of continuous and pulsed exposures on BaP metabolism and antioxidant responses, which were depend on exposure duration.

Seeing Is Believing: Brain-Inspired Modular Training for Mechanistic Interpretability.

We introduce Brain-Inspired Modular Training (BIMT), a method for making neural networks more modular and interpretable. Inspired by brains, BIMT embeds neurons in a geometric space and augments the loss function with a cost proportional to the length of each neuron connection. This is inspired by the idea of minimum connection cost in evolutionary biology, but we are the first the combine this idea with training neural networks with gradient descent for interpretability. We demonstrate that BIMT discovers useful modular neural networks for many simple tasks, revealing compositional structures in symbolic formulas, interpretable decision boundaries and features for classification, and mathematical structure in algorithmic datasets. Qualitatively, BIMT-trained networks have modules readily identifiable by the naked eye, but regularly trained networks seem much more complicated. Quantitatively, we use Newman's method to compute the modularity of network graphs; BIMT achieves the highest modularity for all our test problems. A promising and ambitious future direction is to apply the proposed method to understand large models for vision, language, and science.

Precision Machine Learning.

We explore unique considerations involved in fitting machine learning (ML) models to data with very high precision, as is often required for science applications. We empirically compare various function approximation methods and study how they scale with increasing parameters and data. We find that neural networks (NNs) can often outperform classical approximation methods on high-dimensional examples, by (we hypothesize) auto-discovering and exploiting modular structures therein. However, neural networks trained with common optimizers are less powerful for low-dimensional cases, which motivates us to study the unique properties of neural network loss landscapes and the corresponding optimization challenges that arise in the high precision regime. To address the optimization issue in low dimensions, we develop training tricks which enable us to train neural networks to extremely low loss, close to the limits allowed by numerical precision.

Machine Learning Hidden Symmetries.

We present an automated method for finding hidden symmetries, defined as symmetries that become manifest only in a new coordinate system that must be discovered. Its core idea is to quantify asymmetry as violation of certain partial differential equations, and to numerically minimize such violation over the space of all invertible transformations, parametrized as invertible neural networks. For example, our method rediscovers the famous Gullstrand-Painlevé metric that manifests hidden translational symmetry in the Schwarzschild metric of nonrotating black holes, as well as Hamiltonicity, modularity, and other simplifying traits not traditionally viewed as symmetries.

Driving Ability Evaluation and Rehabilitation for People With Alzheimer's Disease and Related Dementias.

Worldwide, it is estimated that around 50 million older adults have Alzheimer's disease and related dementias (ADRD). Cognitive deficits associated with ADRD may affect a driver's perception and decision-making and potentially cause safety concerns. Despite much research, there lacks a comprehensive cognitive evaluation to determine the driving capability of a person with ADRD and it is unclear what are the most effective training and interventions that help to enhance driving performance for these individuals. The purpose of this article is to conduct a comprehensive literature survey to review and summarize studies of driving performance evaluation and intervention for people with ADRD and discuss perspectives for future studies. Although many studies have investigated the correlations between driving behaviors and cognitive performances for people with ADRD, it remains unclear how driving behaviors and cognitive performances are associated with psychophysiological measures. We discussed the need to develop regular driving evaluation and rehabilitation protocol for people with ADRD. We also highlighted the potential benefit to combine driving tests with psychophysiological measures to assist in characterizing personalized cognitive evaluation in the behavioral evaluation process.

A comparison between ultrasound-guided AIIS injection and radiography in the diagnosis of subspine impingement in patients with FAI.

BACKGROUND: Subspine impingement (SSI) does not have effective diagnostic criteria, especially in patients who also have femoroacetabular impingement (FAI). The classification of anterior inferior iliac spine (AIIS) morphology via three-dimensional CT is controversial. PURPOSE: To propose a method for ultrasound-guided AIIS injection as a way to diagnose SSI and evaluate the accuracy of radiography methods, including 3-D CT and MRI, as well as intraoperative findings. METHODS: Patients diagnosed with FAI between September 2020 and December 2021 were evaluated in this prospective study. Those who met the criteria were included in the ultrasound-guided AIIS injection test. Whether the pain was relieved after injection was recorded in the radiology report. Patients who experienced significant relief of the anterior groin pain (more than 50%) after the AIIS injection were considered positive responders. Among these patients, radiography materials, including AIIS morphology as measured by 3-D CT as well as superior capsular oedema on MRI, were compared. The presence of congestion or bruising on the capsule side of the labrum corresponding to the AIIS during hip arthroscopy was recorded. RESULTS: A total of 73 patients with FAI underwent the ultrasound-guided AIIS injection test. Prevalence rates of 13.70% (10/73), 58.90% (43/73), 23.29% (17/73) and 4.11% (3/73) were recorded for Type I, Type IIA, Type IIB and Type III AIISs, respectively. Thirty-six patients had positive responses to injection, and 37 patients had negative responses to injection. None of the patients with Type I, 23 (53.49%) patients with Type IIA, 11 (64.71%) patients with Type IIB and 2 (66.7%) patients with Type III AIISs had positive responses to the injection. A total of 57.14% of patients with Type II or Type III AIIS had positive responses to the injection. The proportions of patients with superior capsular oedema on MRI in the Type I, Type IIA, Type IIB, and Type III AIIS groups was 0, 30.23, 29.41 and 0%, respectively. Among non-Type I AIIS patients, those who reported positive responses to the injection had a higher incidence of superior capsular oedema (38.89% vs. 14.81%, P = 0.036), but they had no significant differences in the proportion of congestion or bruising of the labrum (47.22% vs. 37.04%, P = 0.419). The results showed that no pairs of methods-ultrasound-guided injection, MRI, and intraoperative findings-achieved good consistency (κ = 0.222, κ = 0.098 and κ = - 0.116). CONCLUSIONS: Radiographic methods including 3-D CT and MRI as well as the intraoperative findings of the labrum cannot be considered an accurate and reliable basis for the diagnosis and treatment of SSI in FAI patients. It is suggested that ultrasound-guided AIIS injections be combined with radiography to better diagnose SSI. LEVEL OF EVIDENCE: IV, case series.

Relapsed boyhood tibia polymicrobial osteomyelitis linked to dermatophytosis: a case report.

BACKGROUND: Relapsed childhood polymicrobial osteomyelitis associated with dermatophytosis has not been reported in the literature. CASE PRESENTATION: Here we report on a case of a 45-year-old man who had left tibial osteomyelitis for 29 years, accompanied by skin fungal infection of the ipsilateral heel for 20 years, and underwent a second operation due to recurrence of polymicrobial infection 6 years ago. The patient had a history of injury from a rusty object, which penetrated the anterior skin of the left tibia middle segment causing subsequent bone infection, but was asymptomatic after receiving treatments in 1983. The patient was physically normal until dermatophytosis occurred on the ipsilateral heel skin in 1998. The patient complained that the dermatophytosis was gradually getting worse, and the tibial wound site became itchy, red, and swollen. The left tibial infection resurged in May 2012, leading to the patient receiving debridement and antibiotic treatment. H&E and Gram-stained histology was performed on biopsy specimens of sequestrum and surrounding inflammatory tissue. Tissue culture and microbiology examination confirmed polymicrobial infection with Staphylococcus aureus (S. aureus) and Corynebacterium and a fungus. Additionally, the patient also received potassium permanganate for dermatophytosis when he was admitted into the hospital. CONCLUSIONS: Together with longitudinal follow-up of medical history, surgical findings, histopathological and microbiology culture evidence, we conclude that boyhood tibia polymicrobial osteomyelitis with S. aureus and Corynebacterium occurred in this patient, and the fungal activation of dermatophytosis may have led to osteomyelitis relapse.

Machine Learning Conservation Laws from Trajectories.

We present AI Poincaré, a machine learning algorithm for autodiscovering conserved quantities using trajectory data from unknown dynamical systems. We test it on five Hamiltonian systems, including the gravitational three-body problem, and find that it discovers not only all exactly conserved quantities, but also periodic orbits, phase transitions, and breakdown timescales for approximate conservation laws.

Multilocation and Multiscale Learning Framework with Skip Connection for Fault Diagnosis of Bearing under Complex Working Conditions.

Considering various fault states under severe working conditions, the comprehensive feature extraction from the raw vibration signal is still a challenge for the diagnosis task of rolling bearing. To deal with strong coupling and high nonlinearity of the vibration signal, this article proposes a novel multilocation and multikernel scale learning framework based on deep convolution encoder (DCE) and bidirectional long short-term memory network (BiLSTM). The procedure of the proposed method using a cascade structure is developed in three stages. In the first stage, each parallel branch of the multifeature learning combines the skip connection and the DCE, and uses different size kernels. The multifeature learning network can automatically extract and fuse global and local features from different network depths and time scales of the raw vibration signal. In the second stage, the BiLSTM as the feature protection network is designed to employ the internal calculating data of the forward propagation and backward propagation at the same network propagation node. The feature protection network is used for further mining sensitive and complementary features. In the third stage of bearing diagnosis, the classifier identifies the fault types. Consequently, the proposed network scheme can perform well in generalization capability. The performance of the proposed method is verified on the two kinds of bearing datasets. The diagnostic results demonstrate that the proposed method can diagnose multiple fault types more accurately. Also, the method performs better in load and speed adaptation compared with other intelligent fault classification methods.

Quadriceps strength and knee joint function in patients with severe knee extension contracture following arthroscopic-assisted mini-incision quadricepsplasty.

PURPOSE: To evaluate reserve quadriceps function and improve knee activity in patients with severe knee extension contracture following arthroscopic-assisted mini-incision quadricepsplasty as well as post-operative complications. METHODS: From 2012 to 2019, 32 patients with severe knee extension contractures (less than 45° range of flexion) were treated with an all-arthroscopic release technique. The clinical results, including range of motion (ROM), quadriceps function (quadriceps index, QI), and knee function, were evaluated, and MRI of the healed tendon after partial quadricepsplasty was performed. The patellar track and length during knee flexion were measured on three normal knees under fluoroscopy. Three formalin-fixed lower limbs were used to mimic severely contracted quadriceps to evaluate the extension of the patellar track. RESULTS: The median follow-up time was 2.1 years (1-5 years). The average QI was 92.0 ± 6.2, and the quadriceps muscle strength was increased from 3.28 to 4.72. At the final follow-up, 90% of the patients had no difficulty going upstairs, going downstairs, or rising from a chair. The ROM improved by 25.69 ± 3.6 preoperatively to 105.88 ± 6.6 at the final follow-up (P < 0.001). The open surgery showed that a 2-cm extension could be achieved by partly cutting the quadriceps tendon, and two cuts achieved a total extension of 5.2 ± 0.52 cm. The patellar tracking distance was 7.7 ± 0.43 cm, and the gap between the patella and femur was also reduced. CONCLUSION: Partial quadricepsplasty of the rectus femoris extended the contracted quadriceps and maintained quadriceps strength, allowing for full knee flexion and satisfactory clinical outcomes of knee function with few complications.

Three-dimensional silk fibroin scaffolds enhance the bone formation and angiogenic differentiation of human amniotic mesenchymal stem cells: a biocompatibility analysis.

Silk fibroin (SF) is a fibrous protein with unique mechanical properties, adjustable biodegradation, and the potential to drive differentiation of mesenchymal stem cells (MSCs) along the osteogenic lineage, making SF a promising scaffold material for bone tissue engineering. In this study, hAMSCs were isolated by enzyme digestion and identified by multiple-lineage differentiation. SF scaffold was fabricated by freeze-drying, and the adhesion and proliferation abilities of hAMSCs on scaffolds were determined. Osteoblast differentiation and angiogenesis of hAMSCs on scaffolds were further evaluated, and histological staining of calvarial defects was performed to examine the cocultured scaffolds. We found that hAMSCs expressed the basic surface markers of MSCs. Collagen type I (COL-I) expression was observed on scaffolds cocultured with hAMSCs. The scaffolds potentiated the proliferation of hAMSCs and increased the expression of COL-I in hAMSCs. The scaffolds also enhanced the alkaline phosphatase activity and bone mineralization, and upregulated the expressions of osteogenic-related factors in vitro. The scaffolds also enhanced the angiogenic differentiation of hAMSCs. The cocultured scaffolds increased bone formation in treating critical calvarial defects in mice. This study first demonstrated that the application of 3D SF scaffolds co-cultured with hAMSCs greatly enhanced osteogenic differentiation and angiogenesis of hAMSCs in vitro and in vivo. Thus, 3D SF scaffolds cocultured with hAMSCs may be a better alternative for bone tissue engineering.

The Chinese ACL injury population has a higher proportion of small ACL tibial insertion sizes than Western patients.

PURPOSE: The study purpose is to characterize the sizes of the anterior cruciate ligament (ACL) insertion site and intercondylar notch in Chinese patients undergoing ACL surgery. The findings will provide a reference for individualized clinical treatment of ACL rupture. METHODS: For this study, 137 patients (102 males, 35 females) with an average age of 30.3 ± 9.5 years (range 14-52 years) undergoing ACL reconstruction were included. The tibial ACL insertion site length and width and the intercondylar notch width were measured on MRI and arthroscopically using a ruler. Descriptive statistics of the patients, the distribution of the measurements and the differences between males and females were calculated. RESULTS: The ACL tibial insertion size and intercondylar notch width in Chinese patients with ACL injuries, as obtained by MRI and intra-operatively, exhibited significant individual variability. The tibial ACL insertion site had a mean length of 13.5 ± 2.1 mm and width of 10.9 ± 1.5 mm as measured on MRI and a mean length of 13.3 ± 2.1 mm and width of 11.0 ± 1.6 mm as measured intra-operatively. The mean intercondylar notch width was 15.2 ± 2.4 mm on MRI and the mean length was 15.0 ± 2.5 mm intra-operatively. The inter-rater reliability between MRI and intra-operative measurements confirmed that the two methods were consistent. In 65.7% of individuals, the ACL tibial insertion length was < 14 mm. CONCLUSION: The distribution of tibial footprint size in Chinese patients is different from that in Western populations. There is a higher proportion of subjects with a tibial footprint size < 14 mm among Chinese patients with ACL injury. Therefore, great care should be taken when treating this population with the double-bundle technique or larger graft options. Level of evidence IV.

[Effects of exercises with different durations and intensities on mitochondrial autophagy and FUNDC1 expression in rat skeletal muscles].

The aim of the present study was to investigate the effects of exercises with different durations and intensities on mitochondrial autophagy and FUNDC1 in rat skeletal muscles. Sixty male Sprague-Dawley rats were randomly divided into 2- and 4-week control groups (Con), moderate-intensity exercise groups (M-ex groups, treadmill exercise, 16 m/min, 1 h/d, 6 d/week), and high-intensity exercise groups (Hi-ex groups, treadmill exercise, 35 m/min, 20 min/d, 6 d/week). The bilateral soleus muscles were separated after the intervention, and paraffin sections were prepared for transmission electron microscopy. ELISA method was used to detect the content of citrate synthase (CS). The co-localizations of microtubule-associated protein 1 light chain 3 (LC3)/cytochrome c oxidase IV (COX-IV), FUNDC1/COX-IV and LC3/FUNDC1 were observed by immunofluorescent staining in frozen sections. The skeletal muscle mitochondria were extracted, and the expression of autophagy-related proteins, including AMPKα, p-AMPKα, Unc-51 like kinase 1 (ULK1), FUNDC1, LC3 and p62, were detected by Western blot. The results showed that exercise increased mitochondrial function, i.e. peroxisome proliferator-activated receptor γ co-activator-1α (PGC-1α), COX-I protein expression levels and CS content. There was no difference of mitochondrial function parameters between 2-week M-ex and 2-week Hi-ex groups, while mitochondrial function of 4-weeks Hi-ex group was significantly lower than that of 4-week M-ex group. Under the same exercise intensity, mitochondrial autophagy activation in skeletal muscle of 4-week exercise was higher than that in 2-week exercise group; Under the same duration of exercise, mitochondrial autophagy activation of Hi-ex group was higher than that in M-ex group. Both 2- and 4-week exercise intervention increased LC3/COX-IV, COX-IV/FUNDC1, and FUNDC1/LC3 co-localizations. Exercise increased LC3-II/LC3-I ratio, down-regulated p62 protein expression level, up-regulated FUNDC1, ULK1 protein expression levels and AMPKα phosphorylation, and the changes of these proteins in 4-week Hi-ex group were significantly greater than those in 4-week M-ex group. These results suggest exercise induces mitochondrial autophagy in skeletal muscles, and the activity of autophagy is related to the duration and intensity of exercise. The induction mechanism of exercise may involve the mediation of FUNDC1 expression through AMPK-ULK1 pathway.

Health impacts of cooking fuel choice in rural China.

This study investigated the impact of cooking fuel choice on the health of elderly people, as measured by activities of daily living, using micro survey data from the China Health and Retirement Longitudinal Study 2015. In contrast to previous studies, our focus on activities of daily living allows for a more comprehensive analysis of health outcomes than diagnoses or doctor visits. Propensity score matching and an endogenous switching regression model were used to address potential selection biases. We found a strong and positive effect of using non-solid cooking fuels on an individual's ability to cope with daily activities, with substantially greater effects on female and older respondents. Our results highlight the need to support energy transition in rural households to non-solid fuels for cooking. We also discuss potential policies to facilitate energy transition in rural China.

Host defense peptide LEAP-2 contributes to monocyte/macrophage polarization in barbel steed (Hemibarbus labeo).

The liver-expressed antimicrobial peptide 2 (LEAP-2) plays a vital role in host immunity against pathogenic organisms. In the present study, cDNA of the LEAP-2 gene was cloned and sequenced from the barbel steed (Hemibarbus labeo). The predicted amino acid sequence of the barbel steed LEAP-2 comprises a signal peptide and a prodomain, which is followed by the mature peptide. Sequence analysis revealed that barbel steed LEAP-2 belongs to the fish LEAP-2A cluster and that it is closely related to zebrafish LEAP-2A. We found that barbel steed LEAP-2 transcripts were expressed in a wide range of tissues, with the highest mRNA levels detected in the liver. In response to lipopolysaccharide (LPS) treatment, LEAP-2 was significantly upregulated in the liver, head kidney, spleen, gill, and mid intestine. A chemically synthesized LEAP-2 mature peptide exhibited selective antimicrobial activity against several bacteria in vitro. Moreover, LEAP-2, alone or in combination with LPS or phorbol 12-myristate 13-acetate, strongly induced a pro-inflammatory reaction in barbel steed monocytes/macrophages (MO/MФ), involving the induction of iNOS activity, respiratory burst, and the pro-inflammatory cytokines IFN-γ, TNF-α, and IL-1ß. Collectively, the results of this study indicate the importance of fish LEAP-2 in the M1-type polarization of MO/MΦ.

All-arthroscopic release for treating severe knee extension contractures could improve the knee range of motion and the mid-term functional outcomes.

PURPOSE: To evaluate the safety, feasibility, and effectiveness of an all-arthroscopic technique for the intra- and extraarticular release of severe knee extension contractures. METHODS: From 2012 to 2016, 25 patients with severe knee extension contractures (less than 45° range of flexion) were treated with an all-arthroscopic release technique. The patients underwent intra- and extraarticular arthroscopic release and arthroscopic-assisted mini-incision quadriceps plasty. The post-operative rehabilitation was initiated the first day after the procedures. Comprehensive clinical follow-up evaluations including the range-of-motion (ROM) assessment, the Lysholm score, and the International Knee Documentation Committee (IKDC) score were performed on all patients. RESULTS: The median follow-up time was 28 months (range 12-65 months). The ROM improved from 23.9° ± 7.5° pre-operatively to 105.9° ± 6.5° at the final follow-up (P < 0.001). In addition, the Lysholm score increased from 59.9 ± 5.2 pre-operatively to 89.7 ± 3.3 (P < 0.001). The IKDC score increased from 47.6 ± 3.4 pre-operatively to 91.7 ± 2.4 (P < 0.001). All patients were satisfied with their final ROM and functional outcomes. CONCLUSION: The all-arthroscopic release technique was a safe, feasible and effective method for treating severe knee extension contractures. The severe knee extension contractures may be successfully addressed by the all-arthroscopic release technique during our clinical practice. LEVEL OF EVIDENCE: IV.