Communication-Efficient and Privacy-Preserving Verifiable Aggregation for Federated Learning.

Federated learning is a distributed machine learning framework, which allows users to save data locally for training without sharing data. Users send the trained local model to the server for aggregation. However, untrusted servers may infer users' private information from the provided data and mistakenly execute aggregation protocols to forge aggregation results. In order to ensure the reliability of the federated learning scheme, we must protect the privacy of users' information and ensure the integrity of the aggregation results. This paper proposes an effective secure aggregation verifiable federated learning scheme, which has both high communication efficiency and privacy protection function. The scheme encrypts the gradients with a single mask technology to securely aggregate gradients, thus ensuring that malicious servers cannot deduce users' private information from the provided data. Then the masked gradients are hashed to verify the aggregation results. The experimental results show that our protocol is more suited for bandwidth-constraint and offline-users scenarios.

PLDP-FL: Federated Learning with Personalized Local Differential Privacy.

As a popular machine learning method, federated learning (FL) can effectively solve the issues of data silos and data privacy. However, traditional federated learning schemes cannot provide sufficient privacy protection. Furthermore, most secure federated learning schemes based on local differential privacy (LDP) ignore an important issue: they do not consider each client's differentiated privacy requirements. This paper introduces a perturbation algorithm (PDPM) that satisfies personalized local differential privacy (PLDP), resolving the issue of inadequate or excessive privacy protection for some participants due to the same privacy budget set for all clients. The algorithm enables clients to adjust the privacy parameters according to the sensitivity of their data, thus allowing the scheme to provide personalized privacy protection. To ensure the privacy of the scheme, we have conducted a strict privacy proof and simulated the scheme on both synthetic and real data sets. Experiments have demonstrated that our scheme is successful in producing high-quality models and fulfilling the demands of personalized privacy protection.

Privacy-preserving COVID-19 contact tracing solution based on blockchain.

The COVID-19 pandemic has severely affected daily life and caused a great loss to the global economy. Due to the very urgent need for identifying close contacts of confirmed patients in the current situation, the development of automated contact tracing app for smart devices has attracted more attention all over the world. Compared with expensive manual tracing approach, automated contact tracing apps can offer fast and precise tracing service, however, over-pursing high efficiency would lead to the privacy-leaking issue for app users. By combing with the benign properties (e.g., anonymity, decentralization, and traceability) of blockchain, we propose an efficient privacy-preserving solution in automated tracing scenario. Our main technique is a combination of non-interactive zero-knowledge proof and multi-signature with public key aggregation. By means of aggregating multiple signatures from different contacts at the mutual commitment phase, we only need fewer zero-knowledge proofs to complete the task of identifying contacts. It inherently leads to the benefits of saving storage and consuming less time for running verification algorithm on blockchain. Furthermore, we perform an experimental comparison by timing the execution of signature verification with and without aggregate signature, respectively. It shows that our solution can actually preserve the full-fledged privacy protection property with a lower computational cost.

Combined fibrinogen and neutrophil-lymphocyte ratio as a biomarker in predicting recurrence of giant cell tumor of bone.

Aims: The present study investigated the combined clinical significance of fibrinogen and neutrophil-lymphocyte ratio (F-NLR) in predicting postoperative recurrence of giant cell tumor of bone. Methods: A total of 113 participants were included in this retrospective study to examine the effects of inflammatory factors on postoperative tumor recurrence. Results: The high-score F-NLR group was significantly associated with larger tumor size (p = 0.001), advanced tumor stage (p = 0.018), wide resection (p = 0.004) and greater local recurrence (p = 0.014). Univariate and multivariate survival analyses revealed that F-NLR (p = 0.035) remained an independent factor influencing tumor recurrence rates. Conclusions: This study reveals that the F-NLR score is a promising blood biomarker for predicting giant cell tumor recurrence.

Giant cell tumor of bone (GCT) is predominantly regarded as an intermediate, locally destructive but rarely metastasizing tumor with recurrence potential and peak incidence in the second to fourth decades of life. The treatment of GCT is often difficult due to local recurrence. Therefore, a new indicator is urgently needed to predict postoperative recurrence more accurately in patients with GCT. The preoperative combined fibrinogen and neutrophil­lymphocyte ratio as a predictor of tumor recurrence and prognosis in GCT patients has been assessed. The present study is the first to demonstrate that a grading system based on the fibrinogen and neutrophil­lymphocyte ratio score is an easily determined, meaningful and reproducible biomarker for predicting recurrence in GCT patients.

Biomechanical analysis of locking plates for fixation of metacarpal shaft fractures: A finite element analysis.

BACKGROUND: There appears to be a paucity of knowledge about the biomechanics of locking plates for the fixation of metacarpal shaft fractures. A thorough understanding of the biomechanics of locking plates is needed to apply them correctly, optimize outcomes, and avoid complications. The purpose of this study is to investigate the biomechanics of the fixation of metacarpal fractures using locking plate-screw constructs with different numbers of screws. HYPOTHESIS: The difference in the number of screws in the locking plate influenced the biomechanical outcome of the metacarpal fracture. METHODS: Finite element models of third metacarpal fractures with locking plate-screw constructs were established, and the magnitude and distribution of their stresses and displacements were investigated when a vertical load of 100N was applied. RESULTS: For the metacarpal fracture with a locking plate and screws, the stress in the metacarpal was largely shared by the plate-screw construct. For the plate-screw construct, the stress is concentrated in the area close to the fracture line, and the 6-screw Group has the lowest failure risk since it has the lowest plate stress and the second-lowest screw stress. The implant-bone construct with 8 screws has better biomechanical stability because of minimal displacement, but increased stress on both the metacarpal bone and the screws, leading to increased failure rates. DISCUSSION: The stresses in the metacarpal were mostly shared by the plate-screw constructs and the screws closest to the fracture line were the most likely to break or loosen. For the implant-bone constructs, the locking plate with 2 screws was the most vulnerable to break or loosen, whereas the locking plate with 6 screws was the least likely to break or loosen. The implant-bone construct with 8 screws had better biomechanical stability, but the stresses in both the metacarpal and the screws were increased, which increased the risk of failure. LEVEL OF EVIDENCE: IV, basic science study.

Privacy-Preserving Multi-Class Support Vector Machine Model on Medical Diagnosis.

With the rapid development of machine learning in the medical cloud system, cloud-assisted medical computing provides a concrete platform for remote rapid medical diagnosis services. Support vector machine (SVM), as one of the important algorithms of machine learning, has been widely used in the field of medical diagnosis for its high classification accuracy and efficiency. In some existing schemes, healthcare providers train diagnostic models with SVM algorithms and provide online diagnostic services to doctors. Doctors send the patient's case report to the diagnostic models to obtain the results and assist in clinical diagnosis. However, case report involves patients' privacy, and patients do not want their sensitive information to be leaked. Therefore, the protection of patient's privacy has become an important research direction in the field of online medical diagnosis. In this paper, we propose a privacy-preserving medical diagnosis scheme based on multi-class SVMs. The scheme is based on the distributed two trapdoors public key cryptosystem (DT-PKC) and Boneh-Goh-Nissim (BGN) cryptosystem. We design a secure computing protocol to compute the core process of the SVM classification algorithm. Our scheme can deal with both linearly separable data and nonlinear data while protecting the privacy of user data and support vectors. The results show that our scheme is secure, reliable, scalable with high accuracy.

Updatable privacy-preserving K -nearest neighbor query in location-based s-ervice.

The K -nearest neighbor ( K -NN) query is an important query in location-based service (LBS), which can query the nearest k points to a given point, and provide some convenient services such as interest recommendations. Hence the privacy protection issue of K -NN query has been a popular research area, protecting the information of queries and the queried results, especially in the information era. However, most of existing schemes fail to consider the privacy protection of location points already stored on servers. Or some schemes support no update of location points. In this paper, we present an updatable and privacy-preserving K -NN query scheme to address the above two issues. Concretely, our scheme utilizes the K D-tree ( K -Dimensional tree) to store the location points of data owners in location service provider and encrypts the points with a distributed double-trapdoor public-key cryptosystem. Then, based on the Ciphertext Comparison Protocol and Ciphertext Euclidean Distance Calculation Protocol, our scheme can protect the privacy of location and query contents. Experimental analyses show our proposal supports some new location points for a fixed location service provider. Moreover, the queried results show a high accuracy of more than 95%.

Retrospective Single-Center Comparative Analysis of Ultrasound-Guided Versus Non-ultrasound-Guided Nerve Blocks in Extremity Surgeries.

OBJECTIVES: Although ultrasound-guided peripheral nerve blocks have become increasingly popular, it is arguable at present whether ultrasound guidance is a necessary tool for peripheral nerve blocks. The purpose of this study was to assess the significance of ultrasound-guided peripheral nerve blocks in reducing complications. METHODS: From January 2013 to January 2019, 17,823 patients who underwent peripheral nerve blocks with/without ultrasound guidance were reviewed, recording data on their age, sex, height, weight, American Society of Anesthesiologists, block type, operation type, operation duration, and complications. The patients were divided into 2 groups: 9372 cases with ultrasound guidance (US Group) and 8451 cases without ultrasound guidance (No-US Group). To control selection bias, a total of 16,236 patients were finally included in this study after 1:1 propensity score matching, including 8118 cases in the US Group and 8118 cases in the No-US Group. Outcome measures included local anesthetic systemic toxicity (LAST), nerve injury, incomplete block anesthesia (IBA), local hematoma, and infection at the block site. RESULTS: Baseline characteristics of patients were similar between the two groups (P >.05, respectively). Our study revealed that the incidence of infection was similar between the two groups (P >.05). However, the incidences of LAST, nerve injury, IBA, and local hematoma in the US Group were significantly lower than those in the No-US Group (P <.05, respectively). CONCLUSIONS: Our findings suggest that ultrasound-guided peripheral nerve blocks are associated with fewer complications and higher success rates, compared with the paresthesia approach.

The cancer-related transcription factor Runx2 combined with osteopontin: a novel prognostic biomarker in resected osteosarcoma.

PURPOSE: Osteosarcoma is the most common primary bone cancer in children and young adults. Recent experimental evidence has indicated that Runx2/OPN axis play important roles in the metastasis of osteosarcoma cells. The present study aimed to explore their relationship and prognostic significance in surgically resected osteosarcoma. METHODS: The expression of runt-related transcription factor2(Runx2) and osteopontin (OPN) in clinical specimens from 105 osteosarcoma patients were detected by immunohistochemistry. The correlations between Runx2, OPN, and clinicopathologic data were analyzed by Chi-square (χ2) tests. The prognostic values were determined by univariate and multivariate survival analysis. The accuracy of oncologic outcome prediction was evaluated by receiver-operating characteristics curves. RESULTS: The results showed there is a significant positive correlation between Runx2 and OPN expression at protein levels (P = 0.015). Runx2 and OPN were both independent predictors for overall survival and metastasis-free survival. When Runx2 and OPN were taken into consideration together, the predictive range was extended and the sensitivity was improved, and more significant and better biomarkers for osteosarcoma metastasis and survival. CONCLUSIONS: These results suggest that a combined Runx2/OPN expression could be a valuable independent predictor of tumor metastasis and survival in osteosarcoma patients.

A Anchoring Technique Using Long Bent Rods for Open Reduction of Thoracolumbar Dislocation.

OBJECTIVE: The purpose of this study was to evaluate the effectiveness of a novel surgical open reduction method for thoracolumbar dislocation. MATERIALS AND METHODS: This study included a total of 15 patients of thoracolumbar dislocation. All patients underwent posterior thoracolumbar open reduction and fixation using this technique. Preoperative x-ray, computed tomography (CT), and magnetic resonance imaging were used to evaluate the dislocation. The American Spinal Injury Association grade of spinal cord injury before and afer the operation were record. All patients were followed up for 2 years, and x-ray, CT were used to observe postoperative reduction and fusion. RESULTS: Postoperative x-ray and CT images of all patients indicated good recovery of the spinal sequence, and no neurological deterioration or surgically related complications occurred. All patients were followed up for 2 years, no patients were lost to follow-up. During the follow-up period, x-ray and CT images showed no complications related to internal fixation such as fracture and loosening of screws and rods were found. CONCLUSION: A anchoring technique using long bent rods is a safe, simple, accurate, and easy to replicate and implement method, which can be applied in the surgical treatment for thoracolumbar dislocation.

Pre-operative prognostic nutritional index was associated with recurrence after surgery in giant cell tumor of bone patients.

OBJECTIVES: Giant cell tumors of bone (GCT) are benign with a local recurrence rate of approximately 20-50%. Growing evidence suggests that inflammation plays an important role in tumor formation and progression. Inflammatory biomarkers, including prognostic nutritional index (PNI), neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR), have little data in predicting postoperative recurrence of GCT. METHODS: We retrospectively investigated 105 patients with surgery for GCT between March 2010 and June 2019 at our hospital. Through the analysis of receiver operating characteristics (ROC), the optimal cutoff values of PNI, NLR and PLR were determined. Clinical features between PNI, NLR and PLR were tested with the χ2 test. Univariate and multivariate analyses were applied to identify the prognostic factors. RESULTS: The optimal cut-off points of PNI, NLR and PLR were 48.6, 2.4 and 136.9, respectively. In univariate analysis, PNI, NLR, PLR, tumor size, Campanacci stage were significantly associated with recurrence-free survival (RFS). Cox multivariate regression analysis revealed that the PNI (p = 0.003) and Campanacci stage (p = 0.001) were independent prognostic factors for GCT. CONCLUSIONS: PNI can be regarded as a novel independent prognostic factor for predicting postoperative recurrence in GCT.

3D Finite Element Analysis of the Modular Prosthesis with Tooth Mechanism of the Femoral Shaft.

OBJECTIVES: To evaluate the mechanical properties and provide a theoretical basis of a diaphyseal prosthesis with tooth mechanism using the finite element analysis method from the point of view of biomechanics. METHODS: A 3D digital femur model was generated based on a 28-year-old healthy man's femoral computed tomography (CT) data in Mimics 17.0 and the customized diaphyseal prostheses with/without tooth mechanism were designed in SolidWorks 2016. The 3D femur model after 8 cm osteotomy in the middle of its shaft and the prostheses with/without tooth mechanism was imported into Abaqus 2016 and the finite element analysis models were established. Three biomechanical tests (compression test, torsion test, and 3P-bending test) under broken load were simulated in FEA to evaluate the performance of the prostheses. RESULTS: The stress distributions of the two prostheses were similar and the maximum von Mises stresses placed on them were very close in each test. The maximum von Mises stresses on the prosthesis with tooth mechanism were 31.55, 319.7, and 447.4 MPa, respectively, and those on the prosthesis without tooth mechanism were 26.26, 300.4, and 455.2 MPa, respectively, in the compression, torsion, and 3P-bending tests. The maximum von Mises stresses on them were far below the ultimate tensile strength or ultimate compressive strength of the titanium alloy. CONCLUSIONS: The diaphyseal prosthesis with tooth mechanism is helpful to adjust the rotation of the long bone during operation. Compared with the conventional diaphyseal prosthesis (without tooth mechanism), the diaphyseal prosthesis with tooth mechanism also has a good biomechanical performance and does not increase the risk of prosthetic failure.

[Minimally invasive treatment for osteonecrosis of the femoral head in ARCO stage â ¡ and â ¢ with bioceramic system].

OBJECTIVE: To perfect the theory system of minimally invasive treatment for osteonecrosis of the femoral head (ONFH) with ß tricalcium phosphate (ß-TCP) bioceramic system and evaluate the effectiveness. METHODS: Eighteen New Zealand white rabbits aged 7-8 months were used to establish an animal model to verify the vascularization of porous ß-TCP bioceramic rods. Micro-CT based three-dimensional reconstruction and fluorescence imaging were used to display the new blood vessels at 4, 8, and 12 weeks after operation. The inserting depth, number and diameter of vessels in the encapsulated area were analyzed. Nine pig femoral specimens were randomly divided into 3 groups ( n=3): group A was normal femur; group B had cavity (core decompression channel+spherical bone defect in femoral head); in group C, mixed bioceramic granules were implanted to fill the defect in femoral head, and porous ß-TCP bioceramic rod was implanted into decompression channel. The stiffness and yield load of specimens were analyzed by biomechanical test. A multicenter retrospective study was conducted to analyze 200 patients (232 hips) with femoral head necrosis treated with bioceramic system in 7 hospitals in China between January 2012 and July 2018. There were 145 males and 55 females, with an average age of 42 years (range, 17-76 years). According to the Association Research Circulation Osseous (ARCO) stage, 150 hips were in stage Ⅱ and 82 hips in stage Ⅲ. Postoperative imaging assessment was carried out regularly, and hip function was evaluated by Harris score. The effectiveness of ARCO stage Ⅱ and Ⅲ was also compared. RESULTS: Animal experiments showed that blood vessels could grow into the encapsulated area and penetrate it at 12 weeks. The inserting depth, number and diameter of blood vessels in the encapsulated area gradually increased, and there was significant difference between different time points ( P<0.05). Biomechanical tests showed that the stiffness and yield load of specimens in groups B and C were significantly lower than those in group A, while the yield load in group B were significantly lower than that in group C ( P<0.05). The stiffness in group C was restored to 41.52%±3.96% in group A, and the yield load was restored to 46.14%±7.85%. Clinical study showed that 200 patients were followed up 6-73 months, with an average of 22.7 months. At last follow-up, 12 patients (16 hips) underwent total hip arthroplasty, and the hip survival rate was 93.10%. According to the imaging evaluation, 184 hips (79.31%) were stable and 48 (20.69%) were worse. Harris score (79.3±17.3) was significantly higher than that before operation (57.3±12.0) ( t=18.600, P=0.000). The excellent rate of hip function was 64.22% (149/232). The survival rate of hip joint, imaging score and Harris score of patients in ARCO stage Ⅱ were better than those in ARCO stage Ⅲ ( P<0.05). CONCLUSION: ß-TCP bioceramic system can guide the abundant blood supply of greater trochanter and femoral neck to the femoral head to promote repair; it can partly restore the mechanical properties of the femoral head and neck in the early stage, providing a new minimally invasive hip-preserving method for patients with ONFH, especially for those in early stage.

PPCD: Privacy-preserving clinical decision with cloud support.

With the prosperity of machine learning and cloud computing, meaningful information can be mined from mass electronic medical data which help physicians make proper disease diagnosis for patients. However, using medical data and disease information of patients frequently raise privacy concerns. In this paper, based on single-layer perceptron, we propose a scheme of privacy-preserving clinical decision with cloud support (PPCD), which securely conducts disease model training and prediction for the patient. Each party learns nothing about the other's private information. In PPCD, a lightweight secure multiplication is presented and introduced to improve the model training. Security analysis and experimental results on real data confirm the high accuracy of disease prediction achieved by the proposed PPCD without the risk of privacy disclosure.

[ANALYSIS OF PROSTHESIS-RELATED COMPLICATIONS AFTER EXTENSIBLE SEMI-JOINT PROSTHESIS REPLACEMENT FOR LOWER LIMBS OSTEOSARCOMA IN CHILDREN].

OBJECTIVE: To investigate the causes of the complications and prevention strategy by analyzing occurrence of prosthesis-related complications after extensible semi-joint prosthesis replacement for lower limbs osteosarcoma in children. METHODS: Eleven children with lower limbs osteosarcoma underwent resection of tumor and replacement of the extensible semi-joint prosthesis between May 2006 and October 2012. There were 6 boys and 5 girls, with an average age of 9.3 years (range, 7-12 years). The lesions located at the distal femur in 6 cases, at the proximal femur in 2 cases, and at the proximal tibia in 3 cases. The disease duration was 2-8 months (mean, 3.6 months). According to the Enneking stage, 3 cases were rated as stage II A and 8 cases as stage II B. The pulmonary CT and ECT results showed no pulmonary metastasis or multi spots before operation. All patients received preoperative chemotherapy treatment for 4 times. RESULTS: Primary healing of incision was obtained in 10 cases. Infection occurred in 1 case at 1 week after operation, and was cured after symptomatic treatment. Nine patients received postoperative chemotherapy for 12 times, 2 patients for 2 times and 4 times respectively. One case died of multiple metastasis; in 3 cases of pulmonary metastasis, 2 cases died and 1 case survived after resection of metastatic lesion. Eight survival cases received a follow-up of 25-89 months (mean, 42.5 months). Loosening and dislocation of the proximal femoral prosthesis occurred in 1 case, loosening and subsidence of the distal femoral prosthesis in 1 case, subluxation in 1 case, and retraction in 1 case. The incidence of prosthesis-related complications was 50%. Lengthening operation was performed on 3 cases for 1 time, and on 1 case for 2 times. And 4 cases did not undergo lengthening operation. According to Enneking function evaluation standard after malignant tumor limb-salvage surgery, the results were excellent in 1 case, good in 3, fair in 3, and poor in 1 at last follow-up with an excellent and good rate of 50%. CONCLUSION: The prosthesis-related complications include loosening and subsidence, dislocation, knee instability, and retraction after extensible semi-joint prosthesis replacement for lower limbs osteosarcoma. The prosthesis-related complications can be reduced by the improvement of prosthesis design and manufacture, and the use of intraoperative bone cement, artificial mesh, and postoperative restrictive brace.

Application of lariat lock catch knot suture in the achilles tendon rupture.

The aim of this study was to summarize the clinical experience of repairing the Achilles tendon rupture by lariat lock catch knot suture. Between January 2011 and February, 2014, 32 cases of the Achilles tendon rupture were treated by lariat lock catch knot suture. There were 26 males and 6 females, with the average age of 39 years (range 17-53 years), including 13 left knees and 19 right knees. 29 wounds healed by first intention, and 3 cases who were performed local flap transfer due to necrosis of skin were healed by second intention. Thirty-two cases were followed up 10-25 months (13 months on average). No re-rupture of Achilles tendon or deep infection occurred during follow-up period. According to Arner-Lindholm assessment standard, the results were excellent in 19 cases and good in 13 cases, the excellent and good rate was 100%. Lariat lock catch knot suture is a safe and effective method for repairing Achilles tendon.