Evaluating a Potential Commercial Tool for Healthcare Application for People with Dementia

Evaluating a Potential Commercial Tool for Healthcare Application for People with Dementia

Tanvi Banerjee1, Pramod Anantharam1 , William Romine2, Larry Lawhorne3,

Amit Sheth1

1Ohio Center of Excellence in Knowledge-enabled Computing(Kno.e.sis),Wright State University, USA

2Department of Biological Sciences, Wright State University, USA3Boonshoft School of Medicine, Wright State University, USA

2

http://www.technologyreview.com/featuredstory/426968/the-patient-of-the-future/

MIT Technology Review, 2012

The Patient of the Future

3

Through analysis of physical, physiological, and environmental observations, our cellphones could act as an early warning system to detect serious health conditions, and provide actionable information

canary in a coal mine

Empowering Individuals (who are not Larry Smarr!) for their own health

kHealth: knowledge-enabled healthcare

AsthmaDementia Heart Failure Liver Cirrhosis

kHealth Application Areas

4

5

1Alzheimer’s Association description of Alzheimer’s statistics, Available online at: http://www.alz.org/alzheimers_disease_facts_and_figures.asp#quickFacts2 Dementia related facts, Available online at: http://www.cdc.gov/mentalhealth/basics/mental-illness/dementia.htm3. K. Vincent, V. A. Velkof, “The next four decades: The older population in the United States: 2010 to 2050.” Washington, D.C.: U.S. Census Bureau; 2010.

5 million

$150 billion

500,000

17.7 billion

People in the U.S. are diagnosed with Alzheimer’s disease1.

Spent on Alzheimer’s alone in a year2

Cause of death in Americans annually

In 2013, hours of unpaid care provided by friends and caregivers3

Dementia: Severity of the problem

6

Public level Signals

Population level Signals

Monitoring and Predicting Behavior Patterns in Patients with Dementia

7

Clinical Collaborators

Dr. Larry Lawhorne, MD

Hexoskin Vest

8

● Heart Rate (HR)● Breathing Rate (BR)● Minute Ventilation (MV)● Cadence● Activity

http://www.hexoskin.com/blogs/news/13591246-hexoskin-wins-most-innovative-consumer-health-product-award-at-interface-future-of-health

Sample Data from a Run Sequence Using the Hexoskin Vest

9

• Test for activity states that can use some known information– Cadence

• Four healthy young subjects completed four activity states (rest, walk, run, and sprint) 10 mins sit10 mins walk10 mins run1 min sprint

Experimental Design: Controlled Study

Activity State Mean Std. Dev

Rest 0.00 0.00

Walk 103.05 25.03

Run 171.95 10.25

Sprint 185.93 22.00

Cadence Validation Across Subjects and Activity States

Key Question: ● What is the consistency of cadence measures across subjects and activity

levels?

Key Assumption:We treat subject and activity state as random effects → attempt

to generalize across all possible subjects and activity states.

Error Analysis: Variance Components Modeling

Effect Estimate % Variance

Subject 133.89 1.78

Activity 7199.19 95.51

Subject-by-Activity 153.91 2.04

Error 50.67 0.67

Results from the Generalizability Study

• Six subjects (increased age range 27 to 68 to include more older adults)

• Longer study: wore the vest for a minimum of two hours• Condition: At least one gait related activity (for cadence)

Experimental Design: Semi-controlled Study

MANOVA Lambda F* R Sq.

Subject 1 0.128 28922.56 0.871

Subject 2 0.160 26888.12 0.839

Subject 3 0.181 32369.65 0.818

Subject 4 0.255 3275.61 0.744

Subject 5 0.375 8020.30 0.624

Subject 6 0.242 6354.81 0.757

MANOVA: Trying to Run multiple regressions on HR, BR, A, MV as DV and C as IV

F critical is 5.1337 at α=.0001

Mean Std. Dev SE Tdf=5 P-value

C-BR 0.54 0.20 0.08 6.53 0.001*

C-HR 0.16 0.28 0.12 1.38 0.226

C-MV 0.66 0.15 0.06 10.9 0.000*

C-A 0.85 0.07 0.03 28.9 0.000*

BR-HR 0.18 0.28 0.11 1.56 0.180

BR-MV 0.18 0.21 0.09 2.04 0.097

BR-A 0.52 0.18 0.07 7.06 0.001*

MV-HR 0.31 0.28 0.11 2.75 0.040*

MV-A 0.64 0.18 0.07 8.93 0.000*

HR-A 0.19 0.28 0.11 1.69 0.152

*Significant at alpha = 0.05

● Cadence is a highly precise indicator of activity states for our cohort ○ Can therefore be used to detect changes in activity patterns across any

individual● Very little individual-level variation in cadence

○ While expected individual effects exist, they are not likely to confound detection of activity changes

● HR was the least correlated with the other variables

Conclusions

Future Work

Carry out a Large Scale Pilot & Clinical Trial• kHealth kit is prepared to be deployed with over 20 or more

dementia patients

Formulate Prediction of Patient’s dementia symptoms using physiological markers from the vest• Personalization is crucial in such a multispectral condition

Add New Sensors for Monitoring sleep and caregiver stress• We need these sensors for caregiver stress with dementia

episodes in patients

Acknowledgements

Partial support for this research was provided by Wright State University’s VP of Research under a challenge grant.

Thank you

Thank you, and please visit us at http://knoesis.org

For more information on kHealth, please visit us at http://knoesis.org/projects/khealth

Link to the paper: http://www.knoesis.org/library/resource.php?id=2155