Dalarna University's logo and link to the university's website

The goal of this study was to develop an algorithm that automatically quantifies motor states (off,on,dyskinesia) in Parkinson's disease (PD), based on accelerometry during a hand pronation-supination test. Clinician's ratings using the Treatment Response Scale (TRS), ranging from -3 (very Off) to 0 (On) to +3 (very dyskinetic), was used as target. For that purpose, 19 participants with advanced PD and 22 healthy persons were recruited in a single center open label clinical trial in Uppsala, Sweden. The trial consisted of single levodopa dose experiments for the people with PD (PwP), where participants were asked to perform standardized wrist rotation tests, using each hand, before and at pre-specified time points after the dose. The participants used wrist sensors containing a 3D accelerometer and gyroscope. Features to quantify the level, variation and asymmetry of the sensor signals, three-level Discrete Wavelet Transform features and approximate entropy measures were extracted from the sensors data. At the time of the tests, the PwP were video recorded. Three movement disorder specialists rated the participants’ state on the TRS scale. A Treatment Response Index from Sensors (TRIS) was constructed to quantify the motor states based on the wrist rotation tests. Different machine learning algorithms were evaluated to map the features derived from the sensor data to the ratings provided by the three specialists. Results from cross validation, both in 10-fold and a leave-one-individual out setting, showed good predictive power of a support vector machine model and high correlation to the TRS scale. Values at the end tails of the TRS scale were under and over predicted due to the lack of observations at those values but the model managed to accurately capture the dose - effect profiles of the patients. In addition, the TRIS had good test-retest reliability on the baseline levels of the PD participants (Intraclass correlation coefficient of 0.83) and reasonable sensitivity to levodopa treatment (0.33 for the TRIS). For a series of test occasions the proposed algorithms provided dose - effect time profiles for participants with PD, which could be useful during therapy individualization of people suffering from advanced PD

Background and Objective

To achieve optimal effect with continuous infusion treatment in Parkinson’s disease (PD), the individual doses (morning dose and continuous infusion rate) are titrated by trained medical personnel. This study describes an algorithmic method to derive optimized dosing suggestions for infusion treatment of PD, by fitting individual dose-response models. The feasibility of the proposed method was investigated using patient chart data.

Methods

Patient records were collected at Uppsala University hospital which provided dosing information and dose-response evaluations. Mathematical optimization was used to fit individual patient models using the records’ information, by minimizing an objective function. The individual models were passed to a dose optimization algorithm, which derived an optimized dosing suggestion for each patient model.

Results

Using data from a single day’s admission the algorithm showed great ability to fit appropriate individual patient models and derive optimized doses. The infusion rate dosing suggestions had 0.88 correlation and 10% absolute mean relative error compared to the optimal doses as determined by the hospital’s treating team. The morning dose suggestions were consistency lower that the optimal morning doses, which could be attributed to different dosing strategies and/or lack of on-off evaluations in the morning.

Conclusion

The proposed method showed promise and could be applied in clinical practice, to provide the hospital personnel with additional information when making dose adjustment decisions.

OBJECTIVE: Dosing schedules for oral levodopa in advanced stages of Parkinson's disease (PD) require careful tailoring to fit the needs of each patient. This study proposes a dosing algorithm for oral administration of levodopa and evaluates its integration into a sensor-based dosing system (SBDS).

MATERIALS AND METHODS: In collaboration with two movement disorder experts a knowledge-driven, simulation based algorithm was designed and integrated into a SBDS. The SBDS uses data from wearable sensors to fit individual patient models, which are then used as input to the dosing algorithm. To access the feasibility of using the SBDS in clinical practice its performance was evaluated during a clinical experiment where dosing optimization of oral levodopa was explored. The supervising neurologist made dosing adjustments based on data from the Parkinson's KinetiGraph™ (PKG) that the patients wore for a week in a free living setting. The dosing suggestions of the SBDS were compared with the PKG-guided adjustments.

RESULTS: The SBDS maintenance and morning dosing suggestions had a Pearson's correlation of 0.80 and 0.95 (with mean relative errors of 21% and 12.5%), to the PKG-guided dosing adjustments. Paired t test indicated no statistical differences between the algorithmic suggestions and the clinician's adjustments.

CONCLUSION: This study shows that it is possible to use algorithmic sensor-based dosing adjustments to optimize treatment with oral medication for PD patients.

The objective of this study is to investigate the effects of feature selection methods on the performance of machine learning methods for quantifying motor symptoms of Parkinson's disease (PD) patients. Different feature selection methods including step-wise regression, Lasso regression and Principal Component Analysis (PCA) were applied on 88 spatiotemporal features that were extracted from motion sensors during hand rotation tests. The selected features were then used in support vector machines (SVM), decision trees (DT), linear regression, and random forests models to calculate a so-called treatment-response index (TRIS). The validity, testretest reliability and sensitivity to treatment were assessed for each combination (feature selection method plus machine learning method). There were improvements in correlation coefficients and root mean squared error (RMSE) for all the machine learning methods, except DTs, when using the selected features from step-wise regression inputs. Using step-wise regression and SVM was found to have better sensitivity to treatment and higher correlation to clinical ratings on the Unified PD Rating Scale as compared to the combination of PCA and SVM. When assessing the ability of the machine learning methods to discriminate between tests performed by PD patients and healthy controls the results were mixed. These results suggest that the choice of feature selection methods is crucial when working with data-driven modelling. Based on our findings the step-wise regression can be considered as the method with the best performance.

OBJECTIVE: The aim of this retrospective study was to investigate if patients with PD, who are treated with levodopa-carbidopa intestinal gel (LCIG), clinically worsen during the afternoon hours and if so, to evaluate whether this occurs in all LCIG-treated patients or in a sub-group of patients.

METHODS: Three published studies were identified and included in the analysis. All studies provided individual response data assessed on the treatment response scale (TRS) and patients were treated with continuous LCIG. Ninety-eight patients from the three studies fulfilled the criteria. T-tests were performed to find differences on the TRS values between the morning and the afternoon hours, linear mixed effect models were fitted on the afternoon hours' evaluations to find trends of wearing-off, and patients were classified into three TRS categories (meaningful increase in TRS, meaningful decrease in TRS, non -meaningful increase or decrease).

RESULTS: In all three studies significant statistical differences were found between the morning TRS values and the afternoon TRS values (p-value <= 0.001 in all studies). The linear mixed effect models had significant negative coefficients for time in two studies, and 48 out of 98 patients (49%) showed a meaningful decrease of TRS during the afternoon hours.

CONCLUSION: The results from all studies were consistent, both in the proportion of patients in the three groups and the value of TRS decrease in the afternoon hours. Based on these findings there seems to be a group of patients with predictable "off" behavior in the later parts of the day. This article is protected by copyright. All rights reserved.