Judges’ Queries and Presenter’s Replies

  • May 20, 2013 | 06:33 p.m.

    What have you learned about Parkinson’s as a result of the monitoring, or what do you think can be learned?

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 20, 2013 | 10:12 p.m.

    Hi Eileen, thank you for your question. I have learned that motor complications of Parkinson’s, such as dyskinesia, can be quantified and identified from body-worn sensors. Within our sample population, the reliability of the system is similar to that of a clinician. This finding has a great implication for people with Parkinson’s: they can automatically have their symptoms monitored by daily wearing a watch-shaped sensor, without requiring personal monitoring from their clinician.

    On a more personal note, I’ve really enjoyed getting to know the Parkinson’s volunteers in my studies! They are a highly motivated group of people, who strive for a better treatment or cure. Most volunteers are active in the community and are eager to give back, despite their impairments. I often find they challenge me to be a better person.

  • May 21, 2013 | 08:54 a.m.

    my question is related to that of Eileen: is there a way for you to “quantify” the value of your research? In other words, can you link your results (current, or those you hope to have by the time you complete this project), with increase in treatment efficiency?
    more important, do you think that data from a set of participants (even if a very large one) could be used for treatment of all/most patients, or would one need to take hese measures for each patient (periodically) in order to provide beter care?

  • May 21, 2013 | 03:49 p.m.

    What additional data would you like to collect assuming that the accelerometer works the way you hope it does? Can you think of a way in which medication could be automatically delivered in response to data collected from the watch? How might the dose be tailored in response to the additional data you would like to collect?

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 21, 2013 | 10:54 p.m.

    Kristopher, in our latest study, we are collecting periodic data that indicates off periods, mood, drowsiness, mental capability, and gait. We are interested in our ability to identify off periods, as well as fluctuating non-motor symptoms such as depression/mood and mental clarity, which greatly impact the quality of life in Parkinson’s. Clinicians treat nonmotor symptoms different than motor symptoms, so a mood tracking system would be valuable to them.

    Regarding automatic medication dosing, probably the easiest way to do this is not through medication but through deep brain stimulation (DBS) settings such as voltage. DBS can induce dyskinesia when the voltage is too high, but the correct amount of stimulation can reduce tremor and other motor complications. The dyskinesia monitoring system could potentially act as a feedback control to adjust DBS voltage within an acceptable range.

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 21, 2013 | 10:54 p.m.

    Ayelet, Yes, it is possible to conduct a study that would quantify the efficacy of this system coupled with clinical treatment in reducing dyskinesia. Because the clinician is still an integral part of the treatment and would only use the system to inform them of patients’ symptoms, the study would need a control group that is treated by clinicians without access to information from the dyskinesia monitoring system. Also, because motor symptoms and complications fluctuate in Parkinson’s, the study would need to be conducted over several months or years to draw conclusions with greater certainty.

    Regarding your second question, this is a really great question we have discussed at length with neurologists. Because each person displays different forms of dyskinesia, and variations also occur over the progression of the disease, the system would need to be “robust” in the sense of identifying all possible types of dyskinesia and classifying to the type that is being presented at the time of sensor observation. Our sample size of 19 participants is far too small to establish a power factor across the entire Parkinson’s population. Likewise, we have not conducted a study that tracks dyskinesia changes with time within individuals. However, the identification accuracy we reported on the poster is across all participants in the study, and is quite high. Preliminarily, we anticipate the system may be able to generalize across the entire Parkinson’s population if it was trained on a large and diverse enough sample population. One of the reasons we think this is because we are calculating a set of 20 features per sensor signal that are designed to capture the critical aspects of different types of dyskinesia reported in the literature. The machine learning algorithm has the ability to select the most important features in identifying dyskinesia across the sample population used to train the algorithm.

  • May 21, 2013 | 09:46 p.m.

    Interesting! I am interested in how you calibrated the decision tree classifier system. How many parameters does the system use, and did you have to initialize them? If so, what data did you use?

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 21, 2013 | 11:11 p.m.

    Timothy,
    The decision tree classifier classifies 20 features per sensor (total of 100 features per participant wearing 5 sensors) into dyskinesia and nondyskinesia periods using visually observed dyskinesia periods as the target class. The decision tree trains on the data set, then validates using 10-fold cross validation. Each instance of 100 features are derived from 1 minute moving time windows of accelerometer data, so there is one instance per 1 minute window. All instances from all participants were included in the training and validation. For a nominal 1.5 hour observation period per participant, this totals about 102600 instances. The tree build nodes based on which feature reduced the classification entropy the most at the level of the node.

  • May 22, 2013 | 09:31 p.m.

    Excellent response! Thank you.

  • Icon for: Mary Gauvain

    Mary Gauvain

    Judge
    May 21, 2013 | 10:59 p.m.

    Is it possible in the course of your research to assess the utility of this technique in relation to the currently used subjective report measure to understand their relation to each other as well as the validity of the subjective report measure? What do you predict? If the two sets of findings are inconsistent, what are the implications? On the other hand, if they are consistent, how would this impact your research?

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 22, 2013 | 12:29 a.m.

    Mary, this is an insightful question. It is interesting because we are currently using observed dyskinesia as our “gold standard” to train our algorithms, even though inter- and intra- rater reliabilities are variable and well documented. When we use human-observed dyskinesia in the algorithm training, we would not expect our algorithms to have greater reliability than the human observers. You ask, is it possible to have greater greater reliability with the machine learning system? Yes, I suspect it is because the system is identifying quantifiable features (such a peak frequencies, energies, amplitudes, etc.) that are documented in the literature as describing dyskinesia, some features of which may not be detectable by human observers. To further investigate this possibility, we have plans to conduct cluster analysis, in which the machine learning algorithms groups features into clusters based on similarities in the data. Although the algorithm would not identify the clusters as belonging to dyskinesia or not, if we could correlate known dyskinesia periods to data within a cluster we could identify that cluster with dyskinesia. This method could potentially surpass the reliability of human observers.

  • Further posting is closed as the competition has ended.

Presentation Discussion

  • May 21, 2013 | 07:52 a.m.

    Great work! Best of luck in the competition. Are you aiming for predicting the upcoming episodes of dyskinesia (Dk)throughout the day in a given day? or are you trying to estimate the overall severity of Dk for any given patient, within a long time period? Are there specific triggers during the day that you know of? or is it totally random when these episodes appear in a given day?

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 21, 2013 | 06:42 p.m.

    Elizabeth, in my most recent study I am identifying both dyskinesia and off periods in Parkinson’s over a one week period. Clinicians have identified knowing the duration of dyskinesia and off periods as important in prescribing treatment such as medication or deep brain stimulation settings. They can optimize treatment to reduce both off & dyskinesia. Therefore I will be providing a timeline graph of dyskinesia and off periods which a clinician could use to optimize treatment. This is more objective than the current method of asking patients to self report over the past several months.
    Triggers to dyskinesia include peak dose carbidopa levodopa medication or too high a voltage on a deep brain stimulation device.

  • May 21, 2013 | 07:49 p.m.

    Thank you Nathan. Your work is extremely valuable and I certainly hope that you win the award. This research is much needed! Also I hope that in a controlled setting your methods eventually aim at separating dyskinesias that are inherent to the disease from those which are induced by medication. The reason for this is because they seem to appear in other populations as well (e.g. autism spectrum disorders and schizophrenia, tardive dyskinesias I think they call them) and it is hard to distinguish medication after-effects from inherent motor problems (which they also have even when no medication is used that could trigger dyskinesias). These things are not well known out there outside the realm of PD, mental illnesses, etc. Yet the need for objective metrics is so high that we should all make an effort to develop good metric. I really commend your work. I think that what you have already achieved is very impressive. Best of luck!

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 21, 2013 | 10:22 p.m.

    Thank you, Elizabeth. Your support is very encouraging. While we have not focused on dyskinesias other than levodopa induced dyskinesia (LID), it seems plausible that other types of dyskinesia would present similarly to LID, and that the system I developed could be trained to detect them. Thank you for bringing my attention to this need!

  • May 21, 2013 | 10:46 p.m.

    You may want to send your methods to the Movement Disorders Society Annual Meeting and they will appreciate them very much. It is something really needed all over the spectrum of PD and mental illnesses. Their membership is a bit expensive to keep up but the field is really eye opening. Engineers like you with your skills are really in high demand because they need so much the types of objective metrics that you are developing here. And I find that the Neurologists are very friendly and open to share their knowledge.

  • Icon for: Nathan Darnall

    Nathan Darnall

    Presenter
    May 22, 2013 | 12:32 a.m.

    Great idea – I’m already a MDS member, and have been collaborating with neurologists in developing the system. Broader input form neurologists would be insightful.

  • Small default profile

    Louise Australia;

    Guest
    May 30, 2013 | 07:26 p.m.

    Knowing people like yourself are involved is immensely comforting and helps keep the flame of hope alive. Thank you for your hard work.

  • Small default profile

    Jillian Carson

    Guest
    May 30, 2013 | 09:27 p.m.

    You need to register in www.parkinsonsummerschool.com taking place in Montreal Canada Sept 15-26 ,2013 right before the 3rd World Parkinson’s congress , take at look at the website

  • Small default profile

    Robert Litowitz

    Guest
    May 31, 2013 | 10:40 a.m.

    Nathan:

    Thank you for your inspiring research. My mom suffered from Parkinson’s. Everything you descrive and show in the video applied to her, and she would have benefitted greatly from your work. Her example inspired my sisters and me to start Parkinsong, which uses music to raise awareness and funding for Parkinson’s Disease. www.parkinsong.com We look forward to learning more about the progress of your study. Thank you and best of luck in the competition

  • Further posting is closed as the competition has ended.

Icon for: Nathan Darnall

NATHAN DARNALL

Washington State University
Years in Grad School: 3
Judges’
Choice

Continuously Identifying Dyskinesia Periods in Parkinson's Disease with Body-Worn Sensors

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that causes both motor and nonmotor deficits, which can be clinically observed and regulated with treatment. Motor symptoms include akinesia, bradykinesia, impaired balance, and tremor. Deficits typically fluctuate in severity on an hourly or daily basis. Time periods characterized by severe deficits are referred to as “off” periods, while periods of relatively normal function are considered “on” periods. Clinicians treat the symptoms of PD through the combined administration of Carbidopa/Levodopa medication, dopamine agonist medication, and voltage-controlled deep brain stimulation (DBS) devices. Overmedication or overstimulation can cause dyskinesia, an involuntary, often rhythmic or choreic, exaggeration of movements. As medication doses are increased with disease progression, dyskinesias become more prevalent. Clinicians aim to reduce both off and dyskinesia periods by prescribing the correct balance of medication and/or DBS treatments. Clinicians investigate the prevalence of dyskinesia, in part, by asking patients to retrospectively self-report the frequency of their dyskinesia periods over the several months prior to their clinical visit, a method subject to recall bias. Our goal in this study was to develop an objective system that identifies dyskinesia periods occurring over a 1-2 hour observational period from body-worn accelerometer data using signal analysis, feature extraction, and machine learning algorithms. We are using the results from this study to design a continuous system that automatically generates a time history of off and dyskinesia periods, which clinicians could use to optimize their prescription of PD treatments.