Using a Motion Sensor to Categorize Nonspecific Low Back Pain Patients: A Machine Learning Approach.

Nonspecific low back pain (NSLBP) constitutes a critical health challenge that impacts millions of people worldwide with devastating health and socioeconomic consequences. In today's clinical settings, practitioners continue to follow conventional guidelines to categorize NSLBP patients based on subjective approaches, such as the STarT Back Screening Tool (SBST). This study aimed to develop a sensor-based machine learning model to classify NSLBP patients into different subgroups according to quantitative kinematic data, i.e., trunk motion and balance-related measures, in conjunction with STarT output. Specifically, inertial measurement units (IMU) were attached to the trunks of ninety-four patients while they performed repetitive trunk flexion/extension movements on a balance board at self-selected pace. Machine learning algorithms (support vector machine (SVM) and multi-layer perceptron (MLP)) were implemented for model development, and SBST results were used as ground truth. The results demonstrated that kinematic data could successfully be used to categorize patients into two main groups: high vs. low-medium risk. Accuracy levels of ~75% and 60% were achieved for SVM and MLP, respectively. Additionally, among a range of variables detailed herein, time-scaled IMU signals yielded the highest accuracy levels (i.e., ~75%). Our findings support the improvement and use of wearable systems in developing diagnostic and prognostic tools for various healthcare applications. This can facilitate development of an improved, cost-effective quantitative NSLBP assessment tool in clinical and home settings towards effective personalized rehabilitation.

Connecting firm's web scraped textual content to body of science: Utilizing microsoft academic graph hierarchical topic modeling.

This paper demonstrates a method to transform and link textual information scraped from companies' websites to the scientific body of knowledge. The method illustrates the benefit of Natural Language Processing (NLP) in creating links between established economic classification systems with novel and agile constructs that new data sources enable. Therefore, we experimented on the European classification of economic activities (known as NACE) on sectoral and company levels. We established a connection with Microsoft Academic Graph hierarchical topic modeling based on companies' website content. Central to the operationalization of our method are a web scraping process, NLP and a data transformation/linkage procedure. The method contains three main steps: data source identification, raw data retrieval, and data preparation and transformation. These steps are applied to two distinct data sources.

Indicators on firm level innovation activities from web scraped data.

This article presents data on companies' innovative behavior measured at the firm-level based on web scraped firm-level data derived from medium-high and high-technology companies in the European Union and the United Kingdom. The data are retrieved from individual company websites and contains in total data on 96,921 companies. The data provide information on various aspects of innovation, most significantly the research and development orientation of the company at the company and product level, the company's collaborative activities, company's products, and use of standards. In addition to the web scraped data, the dataset aggregates a variety firm-level indicators including patenting activities. In total, the dataset includes 21 variables with unique identifiers which enables connecting to other databases such as financial data.

A novel approach to spinal 3-D kinematic assessment using inertial sensors: Towards effective quantitative evaluation of low back pain in clinical settings.

This paper presents a novel approach for evaluating LBP in various settings. The proposed system uses cost-effective inertial sensors, in conjunction with pattern recognition techniques, for identifying sensitive classifiers towards discriminate identification of LB patients. 24 healthy individuals and 28 low back pain patients performed trunk motion tasks in five different directions for validation. Four combinations of these motions were selected based on literature, and the corresponding kinematic data was collected. Upon filtering (4th order, low pass Butterworth filter) and normalizing the data, Principal Component Analysis was used for feature extraction, while Support Vector Machine classifier was applied for data classification. The results reveal that non-linear Kernel classification can be adequately employed for low back pain identification. Our preliminary results demonstrate that using a single inertial sensor placed on the thorax, in conjunction with a relatively simple test protocol, can identify low back pain with an accuracy of 96%, a sensitivity of %100, and specificity of 92%. While our approach shows promising results, further validation in a larger population is required towards using the methodology as a practical quantitative assessment tool for the detection of low back pain in clinical/rehabilitation settings.

A model for flexi-bar to evaluate intervertebral disc and muscle forces in exercises.

This study developed and validated a lumped parameter model for the FLEXI-BAR, a popular training instrument that provides vibration stimulation. The model which can be used in conjunction with musculoskeletal-modeling software for quantitative biomechanical analyses, consists of 3 rigid segments, 2 torsional springs, and 2 torsional dashpots. Two different sets of experiments were conducted to determine the model's key parameters including the stiffness of the springs and the damping ratio of the dashpots. In the first set of experiments, the free vibration of the FLEXI-BAR with an initial displacement at its end was considered, while in the second set, forced oscillations of the bar were studied. The properties of the mechanical elements in the lumped parameter model were derived utilizing a non-linear optimization algorithm which minimized the difference between the model's prediction and the experimental data. The results showed that the model is valid (8% error) and can be used for simulating exercises with the FLEXI-BAR for excitations in the range of the natural frequency. The model was then validated in combination with AnyBody musculoskeletal modeling software, where various lumbar disc, spinal muscles and hand muscles forces were determined during different FLEXI-BAR exercise simulations.

A reliability study of the new sensors for movement analysis (SHARIF-HMIS).

AIM: SHARIF-HMIS is a new inertial sensor designed for movement analysis. The aim of the present study was to assess the inter-tester and intra-tester reliability of some kinematic parameters in different lumbar motions making use of this sensor. MATERIALS AND METHODS: 24 healthy persons and 28 patients with low back pain participated in the current reliability study. The test was performed in five different lumbar motions consisting of lumbar flexion in 0, 15, and 30° in the right and left directions. For measuring inter-tester reliability, all the tests were carried out twice on the same day separately by two physiotherapists. Intra-tester reliability was assessed by reproducing the tests after 3 days by the same physiotherapist. FINDINGS: The present study revealed satisfactory inter- and intra-tester reliability indices in different positions. ICCs for intra-tester reliability ranged from 0.65 to 0.98 and 0.59 to 0.81 for healthy and patient participants, respectively. Also, ICCs for inter-tester reliability ranged from 0.65 to 0.92 for the healthy and 0.65 to 0.87 for patient participants. CONCLUSION: In general, it can be inferred from the results that measuring the kinematic parameters in lumbar movements using inertial sensors enjoys acceptable reliability.

A Mechanical model for flexible exercise bars to study the influence of the initial position of the bar on lumbar discs and muscles forces.

A single-degree-of-freedom model is considered for flexible exercise bars based on the lumped-element approach. By considering the side segment of a flexible bar as a cantilever beam with an equivalent mass at the free end, its free-vibration response, as well as the forced response under the excitation of the grip, are expressed parametrically. Experiments are performed on a particular flexible bar (FLEXI_BAR) in order to obtain numerical values for quantifying the model's parameters. The model is also computationally simulated to study the response of the flexible bar to various excitations. The results are imported into a multi-segment musculoskeletal software (AnyBody), where the effect of different initial hand positions on the lumbar disc and back muscle forces is investigated (including Longissimus, Iliocostalis, and Transversus) during up-down exercises. The results show that all intervertebral discs and muscles forces are more sensitive to the horizontal position of the bar as compared to its vertical position.