Domain Adaptation for Human Activity Recognition of Parkinson's Disease Patients

By Maria Cardei
Slide 1: Title slide for Domain Adaptation for Human Activity Recognition of Parkinson's disease Patients

Slide-1

By: Maria Cardei

Mentor: Behnaz Ghoraani, Ph.D.

FAU I-SENSE REU Scholar 2022

Slide 2: Background information on Human Activity Recognition with smartwatch data collection

Slide-2

Background

  • Human activity recognition (HAR) uses machine learning to classify data acquired by wearable sensors into activities
  • Parkinson's Disease (PD) diagnosis and prognosis can be improved with HAR

a) A smartwatch can collect data for HAR.

Activities shown: Walk, Stand, Sit

b) Poster of Parkinson's Disease Symptoms

Source references: www.micro.ai.com, www.my.clevelandclinic.org

Slide 3: HAR challenges including various sensor types and placements with minimal PD patient data availability

Slide-3

HAR Challenges

  • Various sensor types, placements on body, activity variability between people
  • Minimal PD patient sensor data publicly available

a) Example of sensor placements. Various placements make it difficult to generalize a machine learning model.

b) Machine learning pipeline. A large amount of training data is needed for a successful model.

Source references: www.intellipaat.com, www.breathe.ersjournals.com

Slide 4: Project aim to accurately classify motion data of PD patients into activities of daily living using domain adaptation

Slide-4

Project Aim

Accurately classify the motion data of PD patients into activities of daily living using domain adaptation

Slide 5: Domain Adaptation diagram showing source domain and target domain with feature extractors

Slide-5

Domain Adaptation (DA)

A way to address the domain shift between datasets

The diagram shows:

  • Source Domain with Feature Extractor producing Source Feature
  • Target Domain with Feature Extractor producing Target Feature
  • Similar Classification between the domains

Source reference: https://paperswithcode.com

Slide 6: DANN methodology architecture diagram showing discriminative adversarial neural network components

Slide-6

Methodology - DANN

  • Discriminative Adversarial Neural Network (DANN)
    • Aims to obtain features that are domain invariant
    • Goal is to minimize classification loss (Ly) and maximize discriminator loss (Ld)

a) DANN architecture

Source reference: https://arxiv.org/pdf/1409.7495.pdf

Slide 7: Data methodology showing dataset summary and activity mapping with axes reorientation

Slide-7

Methodology - Data

a) Raw dataset summary before preprocessing

b) Mapping to common activities

Label Healthy Activity PD Activity
0 Standing Standing
1 Sitting Sitting
2 Lying -
3 Walking Walking
4 Climbing stairs -
5 Cycling -
6 Running -

c) Axes before reorientation

Slide 8: Results showing tSNE visualizations comparing DANN model with and without domain adaptation

Slide-8

Results – PAMAP2 to reoriented MHEALTH

Encoded Space tSNE for the Source Only model (Without DA)

Training: PAMAP2 - scatter plot showing source and target

Cross test: reoriented MHEALTH

Source: PAMAP2- scatter plot showing source and target

Target: reoriented MHEALTH

Encoded Space tSNE for the DANN model (With DA)

Slide 9: Performance metrics comparison showing improvement with domain adaptation across accuracy, recall, precision, F1, and AUC

Slide-9

Results – PAMAP2 to reoriented MHEALTH

(averaged over 7 activities) Without DA With DA
Accuracy 48.9% 70.7%
Recall 48.9% 70.7%
Precision 63.8% 64.1%
F1 45.0% 65.1%
AUC 84.0% 95.9%

All metrics improve when DA is applied

Metric definitions:

  • Accuracy = (correct predictions / total predictions) * 100
  • Recall = TP / (TP + FN)
  • Precision = TP / (TP + FP)
  • F1 = harmonic mean of recall and precision
  • AUC = area under ROC curve

Left: two graphs showing Without DA and With DA

Slide 10: Conclusions and future work including DANN validation and next steps for PD data application

Slide-10

Conclusions and Future Work

  • DANN is a valuable DA method
  • Data augmentation exploration
  • Next step is to apply DANN to PD data
    • Challenge of greater domain shift
Slide 11: Questions slide for presentation conclusion

Slide-11

Any questions? Thank you!

Last slide: Contains plain text stating 'End of presentation. Click the right arrow to return to beginning of slide show.'

End of Presentation

Click the right arrow to return to the beginning of the slide show.

For a downloadable version of this presentation, email: I-SENSE@FAU.

Additional Information
The Institute for Sensing and Embedded Network Systems Engineering (I-SENSE) was established in early 2015 to coordinate university-wide activities in the Sensing and Smart Systems pillar of FAU’s Strategic Plan for the Race to Excellence.
Address
Florida Atlantic University
777 Glades Road
Boca Raton, FL 33431
i-sense@fau.edu