Background
Patients with essential tremor (ET), Parkinson’s disease (PD) and dystonic tremor (DT) can be difficult to classify and often share similar characteristics.
Objectives
To use ubiquitous ...smartphone accelerometers with and without clinical features to automate tremor classification using supervised machine learning, and to use unsupervised learning to evaluate if natural clusterings of patients correspond to assigned clinical diagnoses.
Methods
A supervised machine learning classifier was trained to classify 78 tremor patients using leave-one-out cross-validation to estimate performance on unseen accelerometer data. An independent cohort of 27 patients were also studied. Next, we focused on a subset of 48 patients with both smartphone-based tremor measurements and detailed clinical assessment metrics and compared two separate machine learning classifiers trained on these data.
Results
The classifier yielded a total accuracy of 74.4% and F1-score of 0.74 for a trinary classification with an area under the curve of 0.904, average F1-score of 0.94, specificity of 97% and sensitivity of 84% in classifying PD from ET or DT. The algorithm classified ET from non-ET with 88% accuracy, but only classified DT from non-DT with 29% accuracy. A poorer performance was found in the independent cohort. Classifiers trained on accelerometer and clinical data respectively obtained similar results.
Conclusions
Machine learning classifiers achieved a high accuracy of PD, however moderate accuracy of ET, and poor accuracy of DT classification. This underscores the difficulty of using AI to classify some tremors due to lack of specificity in clinical and neuropathological features, reinforcing that they may represent overlapping syndromes.
We sought to test the accuracy of 24-hours ambulatory blood pressure (BP) monitoring (ABPM) for the detection of orthostatic hypotension (OH) in Parkinson's disease (PD). A total of 113 patients ...referred for autonomic testing between January 2015 and June 2017 underwent ABPM and office BP measurements in supine and standing positions. The study population consisted of 81 males and 32 females with PD duration of 6.5 ± 4.1 years and Hoehn and Yahr staging of 1 (13.3%), 1.5 (20.4%), 2 (27.4%), 2.5 (23.9%), 3 (13.3%), and 4 (1.8%). Motor fluctuations were present in 44% of patients. The data from office BP recordings were compared to selected ABPM parameters, and the results showed an association between OH and (a) ABPM-detected hypotensive episodes (Hypo-ep) and (b) ABPM-detected awakening hypotension (Hypo-aw). Having 2 or more Hypo-ep episodes ≤15 mmHg (systolic) compared to average 24-h systolic BP Formula: see text yielded 75% diagnostic accuracy for OH, while the presence of at least one Formula: see text within 90 min after getting up Formula: see text yielded 93% specificity for OH. A diagnostic accuracy of 87.6% was achieved when including daytime and nighttime ABPM values, weighted BP variability, systolic and diastolic BP loads, nocturnal dipping, and postprandial hypotension in a computerized prediction algorithm. In conclusion, our findings suggest that selected ABPM parameters, such as the number of hypotensive episodes and the presence of awakening hypotension, may be used to screen patients for OH, while using a computerized prediction algorithm that includes all ABPM parameters provides the greatest diagnostic accuracy.
Background
Falls represent one of the main complications of Parkinson’s disease (PD), significantly lowering quality of life. Cardiovascular autonomic neuropathy (cAN) is one of the key contributing ...factors to PD-associated falls. However, a direct quantification of its impact on the risk of falling in PD is still lacking. In this 12-month prospective study, we sought to evaluate the association between cAN and falls.
Methods
Fifty consecutive patients were evaluated with a standardized battery of autonomic testing, Unified Parkinson’s Disease Rating Scale, push and release (P&R) test, timed up and go test, freezing of gait (FOG) questionnaire, Montreal cognitive assessment (MoCA). Dyskinesia severity and presence of REM sleep behavioral disorder (RBD) were additionally considered. Patients were followed-up for 12 months.
Results
We observed a 38% prevalence of cAN. At baseline, 36% of patients reported at least one fall in the previous 6 months. This figure increased to 56% over the follow-up. After adjusting for age, disease duration, axial symptoms, MoCA and dopaminergic treatment, cAN was significantly associated with a 15-fold (OR 15.194) higher probability of falls; orthostatic hypotension (OH), the most common expression of cAN, with a 10-fold probability (OR 10.702). In addition P&R test (OR 14.021), RBD (OR 5.470) and FOG (OR 1.450) were independently associated with greater probability of falls.
Conclusions
cAN, including but not limited to OH, is a strong independent predictor of falls in PD. Future research endeavors clarifying to what extent pharmacological and non-pharmacological treatments targeting autonomic dysfunctions might reduce the risk of falls are warranted.
To determine the impact of levodopa-carbidopa intestinal gel (LCIG) infusion on different subtypes of freezing of gait (FoG) classified according to levodopa responsiveness in advanced Parkinson ...disease (PD) patients.
We retrospectively assessed the presence and severity of FoG in 32 advanced PD patients based on the Unified PD Rating Scale (UPDRS) item 14 score. Different FoG subtypes were inferred from the score variation with oral dopaminergic medications. Modifications following long-term LCIG infusion were analysed. Motor symptoms and motor complications were assessed by UPDRS part III and IV respectively.
FoG related UPDRS score varied from 2.6±0.9 in OFF condition to 0.9±0.8 in the ON condition at baseline and improved to 0.6±0.7 with LCIG infusion (p=0.027). After a mean of 2.59±1.12years of continuous LCIG infusion, Pseudo-ON-FoG improved to a greater extent with LCIG infusion than with oral therapy in 12 patients (38%) and equally well in 8 patients (25%), OFF-type-FoG was controlled equally well in 8 patients (25%) and worsened slightly in 3 patients (9%). Unresponsive-FoG, present in one patient (3%), was unmodified by LCIG infusion.
Even though limited by the subjective simple measure of FoG, this study suggests that patients undergoing LCIG infusion maintain a good long-term control of FoG. Pseudo-on-FoG improves in a considerable percentage of patients and OFF-type-FoG remains well controlled with LCIG infusion. Further studies with a larger number of patients and objective measures of FoG are needed to confirm these findings.
•Thirty-two patients treated with LCIG infusion for a mean of 2,5years.•Different FoG types were determined according to levodopa responsiveness.•Pseudo-ON-FoG improved more on LCIG than oral therapy (12 pts) or equally well (8 pts).•OFF-type-FoG improved equally well on LCIG (8 pts) or worsened slightly (3 pts).•LCIG improves or maintains good long-term control of FoG in advanced PD patients.
We sought to assess the effect of subthalamic deep brain stimulation (STN DBS) on Parkinson's disease (PD)-associated postural abnormalities.
A computerized analysis of posture was used to quantify ...the thoracolumbar, thoracic, and cervical-occipital ventral angles, as well as the thoracolumbar and cervical-occipital lateral angles from the video-repository of three specialized movement disorder centers (n = 158 patients). Data was extracted from frames from video-recordings in the pre-surgical medication-ON (dopaminergic therapy) and post-surgical stimulation-ON/medication-ON states (STN DBS plus dopaminergic therapy). The sum of the five postural angles (global postural angle) was used to compare pre-vs. post-surgical trunk posture alterations. A multivariate analysis was used to examine the association between changes in the postural angles and demographic or clinical variables.
There was a 6.7% amelioration in the global postural angle between the pre- and post-surgical assessments (p = 0.031). Motor response to and pre-surgical dosage of levodopa, male gender, and shorter PD duration were identified as predictors for posture improvement after STN DBS. Cases meeting criteria for lower (n = 2) or upper (n = 1) camptocormia respectively improved by 48.1% in the ventral thoracolumbar angle (from 36.4 ± 0.0° to 18.9 ± 4.2°) and 13.8% in the ventral thoracic angle (from 49.1° to 42.3°). Cases meeting criteria for Pisa syndrome (n = 2) improved by 67.5% in the lateral thoracolumbar angle (from 16.9 ± 2.0° to 5.5 ± 4.7°).
STN DBS has a relatively small but significant effect on PD-associated postural abnormalities, potentially enhancing the effect of dopaminergic medications alone.
•Trunk and neck posture alterations represent a major source of disability in PD.•STN DBS showed a beneficial effect on PD-associated posture abnormalities.•STN DBS improved posture in the 5 cases with camptocormia or Pisa syndrome.•Clinical and demographic factors emerged as predictors for posture improvement after STN DBS.