Search for: All records

Multi-instance learning (MIL) handles data that is organized into sets of instances known as bags. Traditionally, MIL is used in the supervised-learning setting for classifying bags which contain any number of instances. However, many traditional MIL algorithms do not scale efficiently to large datasets. In this paper, we present a novel primal–dual multi-instance support vector machine that can operate efficiently on large-scale data. Our method relies on an algorithm derived using a multi-block variation of the alternating direction method of multipliers. The approach presented in this work is able to scale to large-scale data since it avoids iteratively solving quadratic programming problems which are broadly used to optimize MIL algorithms based on SVMs. In addition, we improve our derivation to include an additional optimization designed to avoid solving a least-squares problem in our algorithm, which increases the utility of our approach to handle a large number of features as well as bags. Finally, we derive a kernel extension of our approach to learn nonlinear decision boundaries for enhanced classification capabilities. We apply our approach to both synthetic and real-world multi-instance datasets to illustrate the scalability, promising predictive performance, and interpretability of our proposed method. 

Abstract BackgroundSoft robotic exosuits can provide partial dorsiflexor and plantarflexor support in parallel with paretic muscles to improve poststroke walking capacity. Previous results indicate that baseline walking ability may impact a user’s ability to leverage the exosuit assistance, while the effects on continuous walking, walking stability, and muscle slacking have not been evaluated. Here we evaluated the effects of a portable ankle exosuit during continuous comfortable overground walking in 19 individuals with chronic hemiparesis. We also compared two speed-based subgroups (threshold: 0.93 m/s) to address poststroke heterogeneity. MethodsWe refined a previously developed portable lightweight soft exosuit to support continuous overground walking. We compared five minutes of continuous walking in a laboratory with the exosuit to walking without the exosuit in terms of ground clearance, foot landing and propulsion, as well as the energy cost of transport, walking stability and plantarflexor muscle slacking. ResultsExosuit assistance was associated with improvements in the targeted gait impairments: 22% increase in ground clearance during swing, 5° increase in foot-to-floor angle at initial contact, and 22% increase in the center-of-mass propulsion during push-off. The improvements in propulsion and foot landing contributed to a 6.7% (0.04 m/s) increase in walking speed (R² = 0.82). This enhancement in gait function was achieved without deterioration in muscle effort, stability or cost of transport. Subgroup analyses revealed that all individuals profited from ground clearance support, but slower individuals leveraged plantarflexor assistance to improve propulsion by 35% to walk 13% faster, while faster individuals did not change either. ConclusionsThe immediate restorative benefits of the exosuit presented here underline its promise for rehabilitative gait training in poststroke individuals. 

A quantitative analysis of human gait patterns in space–time provides an opportunity to observe variability within and across individuals of varying motor capabilities. Impaired gait significantly affects independence and quality of life, and thus a large part of clinical research is dedicated to improving gait through rehabilitative therapies. Evaluation of these paradigms relies on understanding the characteristic differences in the kinematics and underlying biomechanics of impaired and unimpaired locomotion, which has motivated quantitative measurement and analysis of the gait cycle. Previous analysis has largely been limited to a statistical comparison of manually selected pointwise metrics identified through expert knowledge. Here, we use a recent statistical-geometric framework, elastic functional data analysis (FDA), to decompose kinematic data into continuous ‘amplitude’ (spatial) and ‘phase’ (temporal) components, which can then be integrated with established dimensionality reduction techniques. We demonstrate the utility of elastic FDA through two unsupervised applications to post-stroke gait datasets. First, we distinguish between unimpaired, paretic and non-paretic gait presentations. Then, we use FDA to reveal robust, interpretable groups of differential response to exosuit assistance. The proposed methods aim to benefit clinical practice for post-stroke gait rehabilitation, and more broadly, to automate the quantitative analysis of motion. 

Multi-instance learning (MIL) is an area of machine learning that handles data that is organized into sets of instances known as bags. Traditionally, MIL is used in the supervised-learning setting and is able to classify bags which can contain any number of instances. This property allows MIL to be naturally applied to solve the problems in a wide variety of real-world applications from computer vision to healthcare. However, many traditional MIL algorithms do not scale efficiently to large datasets. In this paper we present a novel Primal-Dual Multi-Instance Support Vector Machine (pdMISVM) derivation and implementation that can operate efficiently on large scale data. Our method relies on an algorithm derived using a multi-block variation of the alternating direction method of multipliers (ADMM). The approach presented in this work is able to scale to large-scale data since it avoids iteratively solving quadratic programming problems which are generally used to optimize MIL algorithms based on SVMs. In addition, we modify our derivation to include an additional optimization designed to avoid solving a least-squares problem during our algorithm; this optimization increases the utility of our approach to handle a large number of features as well as bags. Finally, we apply our approach to synthetic and real-world multi-instance datasets to illustrate the scalability, promising predictive performance, and interpretability of our proposed method. We end our discussion with an extension of our approach to handle non-linear decision boundaries. Code and data for our methods are available online at: https://github.com/minds-mines/pdMISVM.jl. 

In order to manage the public health crisis associated with COVID-19, it is critically important that healthcare workers can quickly identify high-risk patients in order to provide effective treatment with limited resources. Statistical learning tools have the potential to help predict serious infection early-on in the progression of the disease. However, many of these techniques are unable to take full advantage of temporal data on a per-patient basis as they handle the problem as a single-instance classification. Furthermore, these algorithms rely on complete data to make their predictions. In this work, we present a novel approach to handle the temporal and missing data problems, simultaneously; our proposed Simultaneous Imputation-Multi Instance Support Vector Machine method illustrates how multiple instance learning techniques and low-rank data imputation can be utilized to accurately predict clinical outcomes of COVID-19 patients. We compare our approach against recent methods used to predict outcomes on a public dataset with a cohort of 361 COVID-19 positive patients. In addition to improved prediction performance early on in the progression of the disease, our method identifies a collection of biomarkers associated with the liver, immune system, and blood, that deserve additional study and may provide additional insight into causes of patient mortality due to COVID-19. We publish the source code for our method online.