Machine learning algorithms for activity recognition in ambulant children and adolescents with cerebral palsy

Background

Cerebral palsy (CP) is the most common physical disability among children (2.5 to 3.6 cases per 1000 live births). Inadequate physical activity (PA) is a major problem effecting the health and well-being of children with CP. Practical, yet accurate measures of PA are needed to evaluate the effectiveness of surgical and therapy-based interventions to increase PA. Accelerometer-based motion sensors have become the standard for objectively measuring PA in children and adolescents; however, current methods for estimating physical activity intensity in children with CP are associated with significant error and may dramatically underestimate HPA in children with more severe mobility limitations. Machine learning (ML) models that first classify the PA type and then predict PA intensity or energy expenditure using activity specific regression equations may be more accurate than standalone regression models. However, the feasibility and validity of ML methods has not been explored in youth with CP. Therefore, the purpose of this study was to develop and test ML models for the automatic identification of PA type in ambulant children with CP.

Methods

Twenty two children and adolescents (mean age: 12.8?±?2.9 y) with CP classified at GMFCS Levels I to III completed 7 activity trials while wearing an ActiGraph GT3X+ accelerometer on the hip and wrist. Trials were categorised as sedentary (SED), standing utilitarian movements (SUM), comfortable walking (CW), and brisk walking (BW). Random forest (RF), support vector machine (SVM), and binary decision tree (BDT) classifiers were trained with features extracted from the vector magnitude (VM) of the raw acceleration signal using 10?s non-overlapping windows. Performance was evaluated using leave-one-subject out cross validation.

Results

SVM (82.0–89.0%) and RF (82.6–88.8%) provided significantly better classification accuracy than BDT (76.1–86.2%). Hip (82.7–85.5%) and wrist (76.1–82.6%) classifiers exhibited comparable prediction accuracy, while the combined hip and wrist (86.2–89.0%) classifiers achieved the best overall performance. For all classifiers, recognition accuracy was excellent for SED (94.1–97.9%), good to excellent for SUM (74.0–96.6%) and brisk walking (71.5–86.0%), and modest for comfortable walking (47.6–70.4%). When comfortable and brisk walking were combined into a single walking class, recognition accuracy ranged from 90.3 to 96.5%.

Conclusions

ML methods provided acceptable classification accuracy for detection of a range of activities commonly performed by ambulatory children with CP. The resultant models can help clinicians more effectively monitor bouts of brisk walking in the community. The results indicate that 2-step models that first classify PA type and then predict energy expenditure using activity specific regression equations are worthy of exploration in this patient group.

     

A humanised tissue‐engineered bone model allows species‐specific breast cancer‐related bone metastasis in vivo

Bone metastases frequently occur in the advanced stages of breast cancer. At this stage, the disease is deemed incurable. To date, the mechanisms of breast cancer‐related metastasis to bone are poorly understood. This may be attributed to the lack of appropriate animal models to investigate the complex cancer cell–bone interactions. In this study, two established tissue‐engineered bone constructs (TEBCs) were applied to a breast cancer‐related metastasis model. A cylindrical medical‐grade polycaprolactone‐tricalcium phosphate scaffold produced by fused deposition modelling (scaffold 1) was compared with a tubular calcium phosphate‐coated polycaprolactone scaffold fabricated by solution electrospinning (scaffold 2) for their potential to generate ectopic humanised bone in NOD/SCID mice. While scaffold 1 was found not suitable to generate a sufficient amount of ectopic bone tissue due to poor ectopic integration, scaffold 2 showed excellent integration into the host tissue, leading to bone formation. To mimic breast cancer cell colonisation to the bone, MDA‐MB‐231, SUM1315, and MDA‐MB‐231BO breast cancer cells were cultured in polyethylene glycol‐based hydrogels and implanted adjacent to the TEBCs. Histological analysis indicated that the breast cancer cells induced an osteoclastic reaction in the TEBCs, demonstrating analogies to breast cancer‐related bone metastasis seen in patients.     

Unilateral duplication of the acromioclavicular joint: case report and literature review

Clavicle duplication is a rare anatomical variation of the scapular belt: only seven cases have been reported in the literature to date, and only one took note of the existence of a duplication of the acromioclavicular joint. Two hypotheses have been proposed to interpret this variation: genetic factors, or trauma occurred in the growth period. Clavicle duplication should not be mistaken for a quite frequent coracoclavicular joint widely described. The authors report the case of a left acromioclavicular joint duplication in a 51-year-old male patient presenting with left shoulder pain. This case was the first of literature providing 3D CT-scan images.     

Effects on the buoyant density of rabbit platelets of ADP and agents that increase the concentration of cyclic AMP

Rabbit platelets were aggregated by adenosine diphosphate (ADP), allowed to deaggregate and then separated into density subpopulations by centrifugation through discontinuous Stractan density gradients. Although ADP causes little or no release of the contents of the amine storage granules of rabbit platelets, ADP caused a decrease in platelet density as compared with control platelets subjected to the same procedures except for exposure to ADP. The density change persisted for at least four hours. The apparent size of platelets stimulated with ADP increased initially, but returned to control values during a one-hour period. A similar decrease in platelet density was observed with an albumin density gradient. Under conditions in which aggregation did not occur in response to ADP with ethylenediaminetetraacetic acid (EDTA) in the medium, little or no decrease in platelet density was observed. Agglutination with polylysine did not change platelet density. Thus, not only agents such as thrombin and plasmin that cause the release of the contents of the platelet granules decrease platelet density, but ADP also has this effect. Platelets would be exposed to all of these stimuli during thromboembolic processes, and their effect on platelets may account for the decrease in platelet density observed previously in experiments with rabbits with indwelling aortic catheters. Agents that increase the concentration of cyclic AMP (cAMP) in platelets (PGE1, adenosine, dibutyryl cAMP, forskolin, and papaverine) also decreased platelet density. This effect persisted when the platelets were washed and resuspended in fresh medium and was also demonstrable in plasma. Platelet size was gradually increased by prostaglandin E1 (PGE1) which maintains platelets in a disc shape and does not cause the release of granule contents, indicating that the decrease in platelet density caused by PGE1 may be attributable to platelet swelling.     

Acute toxicity and first clinical results of intensive postinduction therapy using a modified busulfan and cyclophosphamide regimen with autologous bone marrow rescue in first remission of acute myeloid leukemia [see comments]

The combination of high-dose busulfan (16 mg/kg) and 200 mg/kg cyclophosphamide is gaining increasing significance as a preparative regimen prior to autologous, syngeneic, or allogeneic marrow transplantation. A new regimen of high-dose busulfan in conjunction with a reduced dose of 120 mg/kg cyclophosphamide has recently been described as a preparative regimen prior to allogeneic transplantation. To determine the drug-related nonhematologic toxic effects of this new regimen without confounding factors associated with allogeneic transplantation, we conducted a pilot study using this new regimen in 20 patients with acute myeloid leukemia (AML) in first remission prior to autologous unpurged marrow transplantation. All patients experienced transient non-life-threatening acute drug-related toxicity with skin reactions in 20 (100%), nausea and vomiting in 20 (100%), oral mucositis in 18 (90%), hepatic functional impairment in 17 (85%), hemorrhagic cystitis in three (15%), and generalized seizures in two (10%) of these patients, respectively. Two procedural, fatal complications resulted from infectious causes that were not directly related to the speed of hematopoietic reconstitution or the toxicity of the preparative regimen. The 3-year event-free survival estimate (55% +/- 11%) and probability of leukemic recurrence (38% +/- 11%) attained with this new regimen in recipients of autografts in first remission of AML are promising and challenge comparisons with preparative regimens employing combinations of cytotoxic agents or total body irradiation (TBI).     

Immune reconstitution in severe combined immunodeficiency disease after lectin-treated, T-cell-depleted haplocompatible bone marrow transplantation

We describe our 9-year experience with lectin-treated T-cell-depleted haplocompatible parental bone marrow transplantation (BMT) for 24 patients with severe combined immunodeficiency disease (SCID). Nineteen of 21 evaluable patients had T-cell engraftment; 2 of 11 patients tested had B-cell and monocyte engraftment. Fourteen of 24 (58%) patients are alive 7 months to 9.8 years post-BMT. Seventeen of 24 patients received pretransplant conditioning with chemotherapy and/or total body irradiation, and 8 of 24 received more than one transplant. Patients who received conditioning had a survival rate of 61% versus 57% for those who received no conditioning. None received graft-versus- host disease (GVHD) prophylaxis and no patient had acute or chronic GVHD greater than grade I. Kinetics and follow-up of immune recovery were analyzed in 14 patients who are greater than 1 year from transplant. Half of the patients showed evidence of T-cell function by 3 months and normal T-cell function by 4 to 7 months post-BMT. On average, T-cell numbers and subsets became normal 10 to 12 months posttransplant. Recovery of B-cell function was more delayed, although in most patients B-cell numbers and IgM levels were normal by 12 months post-BMT. B-cell function, as determined by isohemagglutinin titers or specific antibodies to pneumococcal polysaccharide, keyhole limpet hemocyanin, or tetanus toxoid, became normal in 10 of 14 patients 2 to 8 years post-BMT. Seven of the 14 are off gammaglobulin therapy. Production of isohemagglutinins tended to predict recovery of antibody response to pneumococcal polysaccharide (P < .064). Based on these results, we believe that haplocompatible BMT is an effective, curative treatment for patients with SCID who lack an HLA-matched related donor.     

牛津单髁膝关节置换：长期结果

牛津膝置换是使用最广泛的膝关节单髁置换（UKR）。牛津膝在37年前开始应用,拥有一个全匹配的活动衬垫,因而磨损率非常低。牛津膝最主要的使用指征是膝关节前内侧骨关节炎,这种病人至少占所有需要行膝关节置换术患者的50％。由于这一系统的设计特点,传统UKR的反指征,如年龄、活动量、肥胖、髌股关节损害和软骨钙质沉着症等对于牛津膝均不是反指征。与全膝关节置换（TKR）相比,牛津膝提供更快的康复、更好的功能、更大的活动度和更好的术后满意度,发生并发症更少、程度更轻,病残率和死亡率更低。一个持续超过30年的研究显示在90％的病例中,牛津膝为患者终生提供了优或良的临床结果,且不需要翻修。在最近15年,牛津膝通过微创手术入路植入,涉及6000多例使用该入路牛津膝置换的9个研究报道显示,10年生存率约95％。在许多这样的研究中,医生们在拟行膝关节置换的患者中约50％使用了牛津单髁膝置换。