Aiming at the problem that the small samples of critical disease in clinic may lead to prognostic models with poor performance of overfitting, large prediction error and instability, the long short-term memory transferring algorithm (transLSTM) was proposed. Based on the idea of transfer learning, the algorithm leverages the correlation between diseases to transfer information of different disease prognostic models, constructs the effictive model of target disease of small samples with the aid of large data of related diseases, hence improves the prediction performance and reduces the requirement for target training sample quantity. The transLSTM algorithm firstly uses the related disease samples to pretrain partial model parameters, and then further adjusts the whole network with the target training samples. The testing results on MIMIC-Ⅲ database showed that compared with traditional LSTM classification algorithm, the transLSTM algorithm had 0.02-0.07 higher AUROC and 0.05-0.14 larger AUPRC, while its number of training iterations was only 39%-64% of the traditional algorithm. The results of application on sepsis revealed that the transLSTM model of only 100 training samples had comparable mortality prediction performance to the traditional model of 250 training samples. In small sample situations, the transLSTM algorithm has significant advantages with higher prediciton accuracy and faster training speed. It realizes the application of transfer learning in the prognostic model of critical disease with small samples.
Speech imagery is an emerging brain-computer interface (BCI) paradigm with potential to provide effective communication for individuals with speech impairments. This study designed a Chinese speech imagery paradigm using three clinically relevant words—“Help me”, “Sit up” and “Turn over”—and collected electroencephalography (EEG) data from 15 healthy subjects. Based on the data, a Channel Attention Multi-Scale Convolutional Neural Network (CAM-Net) decoding algorithm was proposed, which combined multi-scale temporal convolutions with asymmetric spatial convolutions to extract multidimensional EEG features, and incorporated a channel attention mechanism along with a bidirectional long short-term memory network to perform channel weighting and capture temporal dependencies. Experimental results showed that CAM-Net achieved a classification accuracy of 48.54% in the three-class task, outperforming baseline models such as EEGNet and Deep ConvNet, and reached a highest accuracy of 64.17% in the binary classification between “Sit up” and “Turn over”. This work provides a promising approach for future Chinese speech imagery BCI research and applications.
ObjectiveTo compare the short-term outcome of 3-port and 5-port laparoscopic assisted radical resection for middle and high rectal cancer.MethodsWe retrospectively analyzed the clinical characteristics of 67 patients with middle and high rectal cancer who were treated in the Gastrointestinal Ward of Center of General Surgery in General Hospital of Western Theater Command of the Chinese People’s Liberation Army from January 2018 to December 2018. The operative, pathological, recent postoperative related indicators, and follow-up results of the two groups were compared and analyzed.ResultsAmong all the enrolled patients, 33 cases received 3-port laparoscopic surgery (3-port group) and 34 cases received 5-port laparoscopic surgery (5-port group). The total length of incision and the pain score of the Visual Analog Scale (VAS) on the 3rd postoperative day of the 3-port group were significantly better than those of the 5-port group (P<0.05). Peripheral incisal margins were negative in both two groups. However, there were no statistically significant differences in indicators such as operative time, intraoperative blood loss, operative conversion rate, hospitalization expenses, length of the distal margins, number of positive lymph nodes, number of lymph nodes harvested, time to the first flatus, time to the first ambulation, time to urinary catheter removal, time to drainage tube removal, time to the first oral intake, postoperative hospital stay, and postoperative complication rate (P>0.05). Thirty cases of the 3-port group were followed up for 12–24 months (median of 19 months), as well as 29 cases of 5-port group were followed up for 12–24 months (median of 19.5 months). There were no significant differences on the incidences of intestinal obstruction, local recurrence, distant metastasis, and death between the two groups (P>0.05).ConclusionCompared with the conventional 5-port laparoscopic surgery, the 3-port laparoscopic surgery could shorten the total length of incision and reduce the surgical trauma, suggesting that it is safe and effective.
ObjectiveTo investigate the short-term effectiveness and advantages of the orthopedic robot-assisted femoral neck system (FNS) fixation in the treatment of fresh femoral neck fractures compared with the traditional manual operation. Methods A clinical data of 74 patients with fresh femoral neck fractures, who had undergone internal fixation with FNS between April 2020 and September 2021, was retrospectively analyzed. Among them, there were 31 cases of TiRobot-assisted operation (trial group) and 43 cases of traditional manual operation (control group). There was no significant difference between groups (P>0.05) in terms of gender, age, cause of injury, time from injury to operation, fracture side and type. The fracture fixation time (intraoperative fracture reduction to the end of suture), invasive fixation time (incision of internal fixation to the end of suture), the number of placing key-guide needle, incision length, intraoperative blood loss, fracture healing, and Harris score of hip function were recorded and compared. Results All operations were performed with no neurovascular injury or incision complications. The invasive fixation time, intraoperative blood loss, the number of placing key-guide needle, and the incision length in the trial group were superior to the control group (P<0.05), and there was no significant difference in fracture fixation time between groups (P>0.05). All patients were followed up 4-16 months (mean, 7 months). The fracture did not heal in 1 patient of trial group, and the other fractures healed in 2 groups; the fracture healing time was (17.6±1.9) weeks in trial group and (18.2±1.9) weeks in control group, and there was no significant difference between groups (t=0.957, P=0.345). At last follow-up, the Harris score of the trial group was 82.4±5.8, which was higher than that of the control group (79.0±7.7), but the difference was not significant (t=–1.483, P=0.147). Conclusion Orthopedic robot-assisted FNS fixation in the treatment of fresh femoral neck fractures has the similar short-term effectiveness as the traditional method, but the former has advantages in terms of operation time, intraoperative blood loss, and the number of placing key-guide needle, making the operation more minimally invasive and quicker, and more suitable for older patients.
Objective To investigate the effect of body mass index (BMI) on the short-term effectiveness of high tibial osteotomy (HTO) in the treatment of varus knee arthritis. Methods The clinical data of 84 patients (84 knees) with varus knee arthritis treated with HTO between May 2016 and August 2020 were retrospectively analyzed. According to BMI, the patients were divided into normal group (32 patients in group A, BMI<25 kg/m2), overweight group (27 patients in group B, BMI>30 kg/m2), and obese group (25 patients in group C, BMI>30 kg/m2). The BMI of groups A, B, and C were (23.35±0.89), (26.65±1.03), and (32.05±1.47) kg/m2, respectively. There was no significant difference (P>0.05) in gender, age, surgical side, disease duration, and preoperative Hospital for Special Surgery (HSS) score, visual analogue scale (VAS) score, knee range of motion, and hip-knee-ankle angle (HKA) between groups. The operation time, intraoperative dominant blood loss, and the decrease of hemoglobin on the 3rd day after operation were recorded and compared between groups. The improvement of knee joint function and pain status were evaluated by knee joint HSS score, knee range of motion, and VAS score before and after operation, and measuring the HKA of patients on X-ray film. During the follow-up, the X-ray films of the knee joint were reexamined to observe the position of the internal fixator and the healing of osteotomy. Results All patients completed the operation successfully and were followed up 8-40 months (mean, 19.3 months). There was no significant difference in follow-up time, operation time, intraoperative dominant blood loss, and the decrease of hemoglobin on the 3rd day after operation between groups (P>0.05). No operative complications such as severe vascular or nerve injury occurred. After operation, deep venous thrombosis of lower extremities occurred in 1 case in groups A and B respectively, and fat liquefaction of surgical incision occurred in 2 cases in group C. There was no significant difference in the incidence of perioperative complications between groups (3.1% vs. 3.7% vs. 8.0%) (P=0.689). During the follow-up, there was no bone nonunion, plate fracture or loosening. At last follow-up, HSS score, VAS score, knee range of motion, and HKA significantly improved in the 3 groups when compared with those before operation (P<0.05), but there was no significant difference in the differences of the above indexes between groups before and after operation (P>0.05). Conclusion BMI does not affect the short-term effectiveness of HTO in the treatment of varus knee arthritis. HTO can be selected for overweight and obese patients after standard medical treatment is ineffective.
The rapid development of medical informatization and continuous innovation of artificial intelligence have made it possible to analyze data and predict prognosis through making full use of data analysis or data mining methods in medical field, which can provide not only more accurate basis of diagnosis and treatment for patients but also important decision-making reference for the government and hospitals to allocate medical resources reasonably. As a classical model for processing time series data in machine learning, long short-term memory network can break through some limitations of statistics to process large and complex medical data. The current applications of long short-term memory networks in medical and biomedical fields can be mainly summarized as seven themes, including natural language processing, biomedical information, signals, motion, clinical medical records, hospital management, and public health and policy.
Objective To recognize the different phases of Korotkoff sounds through deep learning technology, so as to improve the accuracy of blood pressure measurement in different populations. Methods A classification model of the Korotkoff sounds phases was designed, which fused attention mechanism (Attention), residual network (ResNet) and bidirectional long short-term memory (BiLSTM). First, a single Korotkoff sound signal was extracted from the whole Korotkoff sounds signals beat by beat, and each Korotkoff sound signal was converted into a Mel spectrogram. Then, the local feature extraction of Mel spectrogram was processed by using the Attention mechanism and ResNet network, and BiLSTM network was used to deal with the temporal relations between features, and full-connection layer network was applied in reducing the dimension of features. Finally, the classification was completed by SoftMax function. The dataset used in this study was collected from 44 volunteers (24 females, 20 males with an average age of 36 years), and the model performance was verified using 10-fold cross-validation. Results The classification accuracy of the established model for the 5 types of Korotkoff sounds phases was 93.4%, which was higher than that of other models. Conclusion This study proves that the deep learning method can accurately classify Korotkoff sounds phases, which lays a strong technical foundation for the subsequent design of automatic blood pressure measurement methods based on the classification of the Korotkoff sounds phases.
The pace of modern life is accelerating, the pressure of life is gradually increasing, and the long-term accumulation of mental fatigue poses a threat to health. By analyzing physiological signals and parameters, this paper proposes a method that can identify the state of mental fatigue, which helps to maintain a healthy life. The method proposed in this paper is a new recognition method of psychological fatigue state of electrocardiogram signals based on convolutional neural network and long short-term memory. Firstly, the convolution layer of one-dimensional convolutional neural network model is used to extract local features, the key information is extracted through pooling layer, and some redundant data is removed. Then, the extracted features are used as input to the long short-term memory model to further fuse the ECG features. Finally, by integrating the key information through the full connection layer, the accurate recognition of mental fatigue state is successfully realized. The results show that compared with traditional machine learning algorithms, the proposed method significantly improves the accuracy of mental fatigue recognition to 96.3%, which provides a reliable basis for the early warning and evaluation of mental fatigue.
Objective To investigate the short-term effectiveness of reconstruction plate internal fixation via improved Stoppa approach combined with iliac fossa approach and Kocher-Langenbeck approach for complex acetabular fractures. Methods Between January 2015 and January 2020, 30 patients with complex acetabular fractures were treated with reconstruction plate internal fixation via improved Stoppa approach combined with iliac fossa approach and Kocher-Langenbeck approach. There were 20 males and 10 females with an average age of 52.1 years (range, 25-71 years). The cause of injury included traffic accident in 17 cases and falling from height in 13 cases. Among them, 14 cases were left acetabular fractures and 16 cases were right acetabular fractures. According to Letournel classification, there were 16 cases of double column fractures, 2 cases of transverse fracture with posterior wall fracture, 4 cases of anterior column and posterior hemi-transverse fractures, 8 cases of T-shaped fracture. The displacement distance of fracture ranged from 6 to 30 mm (mean, 11.6 mm). The time from injury to operation was 6-14 days (mean, 8.7 days). Results The operation time was 2.0-4.5 hours (mean, 3.0 hours). The intraoperative blood loss was 200-800 mL (mean, 450.0 mL). All patients were treated with autologous blood transfusion during operation. All incisions healed by first intention after operation, and no infection occurred. All patients were followed up 12-15 months (mean, 13.4 months). The drainage tube was removed at 2-3 days after operation. After extubation, X-ray film and three-dimensional CT were performed to recheck the fracture reduction. According to Matta score system, 20 cases were excellent, 5 were good, and 5 were poor, the excellent and good rate was 83.3%. All fractures healed with the healing time of 16-25 weeks (mean, 17.7 weeks). According to Merle d’Aubigne-Postel score system, the hip function at 1 year after operation was rated as excellent in 18 cases, good in 6 cases, and fair in 6 cases, and the excellent and good rate was 80.0%. Two cases suffered from sciatic nerve injury due to traction during operation, 7 cases had heterotopic ossification, 2 cases had traumatic hip arthritis, and no other complications occurred. Conclusion For complex acetabular fractures, the reconstruction plate internal fixation via improved Stoppa approach combined with iliac fossa approach and Kocher-Langenbeck approach can obtain good short-term effectiveness with good reduction and hip joint function, and less complications.
ObjectiveTo evaluate the safety and short-term outcome of laparoscopic total mesorectal excision (TME) for the middle-lower rectal cancer in municipal hospital.MethodsThe pathological data of 94 patients with middle-lower rectal cancer (49 cases underwent laparoscopic TME, while 45 cases received open TME), who treated in The First People’s Hospital of Ziyang from Jan. 2015 to Jun. 2017, were retrospectively collected and analyzed.ResultsTwo patients (4.1%) in laparoscopy group were converted to open surgery. Compared with the laparotomy group, the laparoscopic group had significantly less volume of intraoperative bleeding, shorter abdominal incision, earlier time to the first flatus and liquid diet, and lower rate of 30-day postoperative complication (P<0.05), but had longer operative time (P=0.033). While there were no significant difference on postoperative stay, the specimen length, distal margin, and number of harvested lymph nodes between the 2 groups (P>0.05).ConclusionLaparoscopic TME is a feasible, safe, and minimally invasive technique for middle-lower rectal cancer, and produce more favourable short-term outcome than open surgery in municipal hospital.