Luca Palmerini

Feature Selection for Accelerometer-Based Posture Analysis in Parkinson's Disease

Posture analysis in quiet standing is a key component of the clinical evaluation of Parkinsons disease (PD), postural instability being one of PDs major symptoms. The aim of this study was to assess the feasibility of using accelerometers to characterize the postural behavior of early mild PD subjects. Twenty PD and 20 control subjects, wearing an accelerometer on the lower back, were tested in five conditions characterized by sensory and attentional perturbation. A total of 175 measures were computed from the signals to quantify tremor, acceleration, and displacement of body sway. Feature selection was implemented to identify the subsets of measures that better characterize the distinctive behavior of PD and control subjects. It was based on different classifiers and on a nested cross validation, to maximize robustness of selection with respect to changes in the training set. Several subsets of three features achieved misclassification rates as low as 5%. Many of them included a tremor-related measure, a postural measure in the frequency domain, and a postural displacement measure. Results suggest that quantitative posture analysis using a single accelerometer and a simple test protocol may provide useful information to characterize early PD subjects. This protocol is potentially usable to monitor the diseases progression.

Quantification of Motor Impairment in Parkinson's Disease Using an Instrumented Timed Up and Go Test

The Timed Up and Go (TUG) test is a clinical test to assess mobility in Parkinsons disease (PD). It consists of rising from a chair, walking, turning, and sitting. Its total duration is the traditional clinical outcome. In this study an instrumented TUG (iTUG) was used to supplement the quantitative information about the TUG performance of PD subjects: a single accelerometer, worn at the lower back, was used to record the acceleration signals during the test and acceleration-derived measures were extracted from the recorded signals. The aim was to select reliable measures to identify and quantify the differences between the motor patterns of healthy and PD subjects; in order to do so, besides comparing each measure individually to find significant group differences, feature selection and classification were used to identify the distinctive motor pattern of PD subjects. A subset of three features (two from Turning, one from the Sit-to-Walk component), combined with an easily-interpretable classifier (Linear Discriminant Analysis), was found to have the best accuracy in discriminating between healthy and early-mild PD subjects. These results suggest that the proposed iTUG can characterize PD motor impairment and, hence, may be used for evaluation, and, prospectively, follow-up, and monitoring of disease progression.

Hilbert–Huang-Based Tremor Removal to Assess Postural Properties From Accelerometers

Tremor is one of the symptoms of several disorders of the central and peripheral nervous system, such as Parkinsons disease (PD). The impairment of postural control is another symptom of PD. The conventional method of posture analysis uses force plates, but accelerometers can be a valid and reliable alternative. Both these measurement techniques are sensitive to tremor. Tremor affects postural measures and may thus lead to misleading results or interpretations. Linear low-pass filters (LPFs) are commonly employed for tremor removal. In this study, an alternative method, based on Hilbert-Huang transformation (HHT), is proposed. We examined 20 PD subjects, with and without tremor, and 20 control subjects. We compared the effectiveness of LPF and HHT-based filtering on a set of postural parameters extracted from acceleration signals. HHT has the advantage of providing a filter, which with no a priori knowledge, efficiently manages the nonlinear, nonstationary interference due to tremor, and beyond tremor, gives descriptive measures of postural function. Some of the differences found using LPF can instead be ascribed to inefficient noise/tremor suppression. Filter order and cutoff frequency are indeed critical when subjects exhibit a tremorous behavior, in which case LPF parameters should be chosen very carefully.

A wavelet-based approach to fall detection

Falls among older people are a widely documented public health problem. Automatic fall detection has recently gained huge importance because it could allow for the immediate communication of falls to medical assistance. The aim of this work is to present a novel wavelet-based approach to fall detection, focusing on the impact phase and using a dataset of real-world falls. Since recorded falls result in a non-stationary signal, a wavelet transform was chosen to examine fall patterns. The idea is to consider the average fall pattern as the “prototype fall”.In order to detect falls, every acceleration signal can be compared to this prototype through wavelet analysis. The similarity of the recorded signal with the prototype fall is a feature that can be used in order to determine the difference between falls and daily activities. The discriminative ability of this feature is evaluated on real-world data. It outperforms other features that are commonly used in fall detection studies, with an Area Under the Curve of 0.918. This result suggests that the proposed wavelet-based feature is promising and future studies could use this feature (in combination with others considering different fall phases) in order to improve the performance of fall detection algorithms.

FRAT-up, a Web-based Fall-Risk Assessment Tool for Elderly People Living in the Community

Background About 30% of people over 65 are subject to at least one unintentional fall a year. Fall prevention protocols and interventions can decrease the number of falls. To be effective, a prevention strategy requires a prior step to evaluate the fall risk of the subjects. Despite extensive research, existing assessment tools for fall risk have been insufficient for predicting falls. Objective The goal of this study is to present a novel web-based fall-risk assessment tool (FRAT-up) and to evaluate its accuracy in predicting falls, within a context of community-dwelling persons aged 65 and up. Methods FRAT-up is based on the assumption that a subject’s fall risk is given by the contribution of their exposure to each of the known fall-risk factors. Many scientific studies have investigated the relationship between falls and risk factors. The majority of these studies adopted statistical approaches, usually providing quantitative information such as odds ratios. FRAT-up exploits these numerical results to compute how each single factor contributes to the overall fall risk. FRAT-up is based on a formal ontology that enlists a number of known risk factors, together with quantitative findings in terms of odds ratios. From such information, an automatic algorithm generates a rule-based probabilistic logic program, that is, a set of rules for each risk factor. The rule-based program takes the health profile of the subject (in terms of exposure to the risk factors) and computes the fall risk. A Web-based interface allows users to input health profiles and to visualize the risk assessment for the given subject. FRAT-up has been evaluated on the InCHIANTI Study dataset, a representative population-based study of older persons living in the Chianti area (Tuscany, Italy). We compared reported falls with predicted ones and computed performance indicators. Results The obtained area under curve of the receiver operating characteristic was 0.642 (95% CI 0.614-0.669), while the Brier score was 0.174. The Hosmer-Lemeshow test indicated statistical significance of miscalibration. Conclusions FRAT-up is a web-based tool for evaluating the fall risk of people aged 65 or up living in the community. Validation results of fall risks computed by FRAT-up show that its performance is comparable to externally validated state-of-the-art tools. A prototype is freely available through a web-based interface. Trial Registration ClinicalTrials.gov NCT01331512 (The InChianti Follow-Up Study); http://clinicaltrials.gov/show/NCT01331512 (Archived by WebCite at http://www.webcitation.org/6UDrrRuaR).

Dimensionality reduction for the quantitative evaluation of a smartphone-based Timed Up and Go test

The Timed Up and Go is a clinical test to assess mobility in the elderly and in Parkinsons disease. Lately instrumented versions of the test are being considered, where inertial sensors assess motion. To improve the pervasiveness, ease of use, and cost, we consider a smartphones accelerometer as the measurement system. Several parameters (usually highly correlated) can be computed from the signals recorded during the test. To avoid redundancy and obtain the features that are most sensitive to the locomotor performance, a dimensionality reduction was performed through principal component analysis (PCA). Forty-nine healthy subjects of different ages were tested. PCA was performed to extract new features (principal components) which are not redundant combinations of the original parameters and account for most of the data variability. They can be useful for exploratory analysis and outlier detection. Then, a reduced set of the original parameters was selected through correlation analysis with the principal components. This set could be recommended for studies based on healthy adults. The proposed procedure could be used as a first-level feature selection in classification studies (i.e. healthy-Parkinsons disease, fallers-non fallers) and could allow, in the future, a complete system for movement analysis to be incorporated in a smartphone.

Fall risk assessment tools for elderly living in the community: can we do better?

Background Falls are a common, serious threat to the health and self-confidence of the elderly. Assessment of fall risk is an important aspect of effective fall prevention programs. Objectives and methods In order to test whether it is possible to outperform current prognostic tools for falls, we analyzed 1010 variables pertaining to mobility collected from 976 elderly subjects (InCHIANTI study). We trained and validated a data-driven model that issues probabilistic predictions about future falls. We benchmarked the model against other fall risk indicators: history of falls, gait speed, Short Physical Performance Battery (Guralnik et al. 1994), and the literature-based fall risk assessment tool FRAT-up (Cattelani et al. 2015). Parsimony in the number of variables included in a tool is often considered a proxy for ease of administration. We studied how constraints on the number of variables affect predictive accuracy. Results The proposed model and FRAT-up both attained the same discriminative ability; the area under the Receiver Operating Characteristic (ROC) curve (AUC) for multiple falls was 0.71. They outperformed the other risk scores, which reported AUCs for multiple falls between 0.64 and 0.65. Thus, it appears that both data-driven and literature-based approaches are better at estimating fall risk than commonly used fall risk indicators. The accuracy–parsimony analysis revealed that tools with a small number of predictors (~1–5) were suboptimal. Increasing the number of variables improved the predictive accuracy, reaching a plateau at ~20–30, which we can consider as the best trade-off between accuracy and parsimony. Obtaining the values of these ~20–30 variables does not compromise usability, since they are usually available in comprehensive geriatric assessments.

A Probabilistic Model to Investigate the Properties of Prognostic Tools for Falls

BACKGROUND Falls are a prevalent and burdensome problem in the elderly. Tools for the assessment of fall risk are fundamental for fall prevention. Clinical studies for the development and evaluation of prognostic tools for falls show high heterogeneity in the settings and in the reported results. Newly developed tools are susceptible to over-optimism. OBJECTIVES This study proposes a probabilistic model to address critical issues about fall prediction through the analysis of the properties of an ideal prognostic tool for falls. METHODS The model assumes that falls occur within a population according to the Greenwood and Yule scheme for accident-proneness. Parameters for the fall rate distribution are estimated from counts of falls of four different epidemiological studies. RESULTS We obtained analytic formulas and quantitative estimates for the predictive and discriminative properties of the ideal prognostic tool. The area under the receiver operating characteristic curve (AUC) ranges between about 0.80 and 0.89 when prediction on any fall is made within a follow-up of one year. Predicting on multiple falls results in higher AUC. CONCLUSIONS The discriminative ability of current validated prognostic tools for falls is sensibly lower than what the proposed ideal perfect tool achieves. A sensitivity analysis of the predictive and discriminative properties of the tool with respect to study settings and fall rate distribution identifies major factors that can account for the high heterogeneity of results observed in the literature.

Balance testing with inertial sensors in patients with Parkinson's disease: assessment of motor subtypes.

In this study, the use of an instrumented balance test based on inertial sensors was evaluated in patients with Parkinsons disease (PD). We aimed to objectively characterize motor subtypes of PD [tremor dominant (TD) and postural instability gait difficulty (PIGD)], to help to quantitatively classify the PD subjects into motor subtypes. Subjects were studied performing postural tests, while wearing a device including a tri-axial accelerometer on the lower back, in four different experimental conditions that depended on feet position (feet-together or semi-tandem) and vision (eyes open or closed). Postural measures, after a reliability check, were tested to identify their sensitivity to the disease, to the PD subtypes, and to the experimental conditions. The results highlight the possibility of distinguishing between the TD and PIGD subtypes by means of objective postural measures that are able to detect tremor and PIGD features and are able to classify a subject as TD or PIGD with good accuracy. Feet position influences frequency measures, whereas eyes closure influences the displacement measures and enhances differences between PD and control subjects, suggesting that postural displacement measures may be capable of detecting different adaptation processes to external sensory conditions between patients with PD and control subjects.

Classification of Early-Mild Subjects with Parkinson’s Disease by Using Sensor-Based Measures of Posture, Gait, and Transitions

Evaluation of posture, gait, turning, and different kind of transitions, are key components of the clinical evaluation of Parkinson’s disease (PD). The aim of this study is to assess the feasibility of using accelerometers to classify early PD subjects (two evaluations over a 1-year follow-up) with respect to age-matched control subjects. Classifying PD subjects in an early stage would permit to obtain a tool able to follow the progression of the disease from the early phases till the last ones and to evaluate the efficacy of different treatments. Two functional tests were instrumented by a single accelerometer (quiet standing, Timed Up and Go test); such tests carry quantitative information about impairments in posture, gait, and transitions (i.e. Sit-to-Walk, and Walk-to-Sit, Turning). Satisfactory accuracies are obtained in the classification of PD subjects by using an ad hoc wrapper feature selection technique.