Technology to Support Individuals with Cognitive Impairment Martha E. Pollack Computer Science & Engineering University of Michigan.

Technology to Support Individuals with Cognitive Impairment

Martha E. PollackComputer Science &

EngineeringUniversity of Michigan

Challenges for Older Adults

• Physical

• Social/emotional

• Sensory

• Cognitive– Example: Alzheimer’s

65-74: 5%75-84: 20%> 85: 50%

intelligent wheelchairs

elder-friendly email and chat rooms

programmable digital hearing aids

Some of the technology also useful for younger individuals with cognitiveimpairment (e.g., TBI patients, developmentally disabled people)

Why Build Cognitive Orthotics?

• Cognitive impairment can impact performance of daily activities

• Can lead to decreased quality of life, and potentially institutionalization– Costly

– Further decreases quality of life

• Goals– Improve performance of routine functional activities and thereby

support longer aging-in-place

– Reduce caregiver burden

Activity Cueing

• Guide an individual through steps in a sequential or conditional-branching process

• Work done both on ADLs/IADLs (e.g., handwashing, cooking) and on functional job tasks (e.g., janitorial)

Handwashing Assistant [courtesy A. Mihailidis, U. Toronto]

Prospective Memory Aids

• Tend to be designed for less severely impaired individuals

• Provide them with personalized, adaptive reminders about daily activities

• On the market: glorified alarm clocks!– Exception: PEAT

Autominder

• Model, update, and maintain the client’s plan– Including complex temporal and causal constraints

• Monitor the client’s performance– Updating the plan as execution proceeds

• Reason about what reminders to issue, and when– To most effectively ensure compliance, without sacrificing

client independence

Autominder Example

Req/Opt Activity Allowed Expected Observed

R toilet use 10:45-11:05

R lunch 12:00-12:45

O TV 14:00-14:30

10:55

R toilet use13:55-14:15

REMIND 12:25

REMIND 13:55

12:28

Robot Platform

• Nomadic Technologies Scout II

w/custom-designed head

– Multiple sensors: lasers, sonars, microphone, touchscreen, camera vision, wireless ethernet

– Effectors: motion, speakers, display screen, facial expression

“Pearl”[courtesy Carnegie MellonUniv. Robotics Institute]

“Ubicomp” Platform

• Handheld or wearable device– Currently: HP iPaq

• Deployed in a “smart” environment with multiple sensors (ubiquitous computing environment)

Client Client ModelerModeler

Plan Plan ManagerManager

IntelligentIntelligentReminderReminderGeneratorGenerator

ClientPlan

Activity Info

Inferred Activity

Sensor Data

Reminders

Client Model Info

Activity Info

Preferences

Plan Updates

ClientModel

Autominder ArchitectureWhat should the client do?

Technologies: Automated Planning, Constraint-Based Temporal Reasoning

What is the client doing?

Technologies: Dynamic Bayesian InferenceIs a reminder needed?

Technologies: Iterative Refinement Planning, Reinforcement Learning

Plan Manager: What should the client do?

• Maintains up-to-date record of client’s planned activities– Eating, hydrating, toileting, medicine-taking, exercise, social activities,

doctor’s appointments, etc.

• Updates plan and propagates constraints when– New planned activity added.

– Existing activity modified or deleted.

– Planned activity performed.

– Critical time bounds passed.

• Models plans as Disjunctive Temporal Problems and uses AI planning and CSP technology for updating.

Client Modeler: What is the client doing?

• Given information:– Sensor input: client moved to kitchen– Clock time: at 7:23 a.m.– Client plan: breakfast should be eaten between 7 and 8

– Model of previous actions: client has not yet eaten breakfast– Learned patterns: 82% of the time, client starts breakfast between 7:10 and

7:25– Reminder information: we issued a reminder at 7:21

• Infers probability that various events have occurred– that the client has begun breakfast

• Uses Bayesian reasoning technology, addressing limitations of previous approaches to handle complex and dynamic temporal relations

Intelligent Reminder Generation: What should Autominder do?

• Given a client’s plan and its execution status:– Easy to generate reminders

• Remind at earliest possible time of each action

– Harder to “remind well”• Maximize likelihood of appropriate performance of ADLs and

other key activities• Facilitate efficient performance• Avoid annoying client• Avoid making client overly reliant

• Uses local search tools to incrementally refine reminder plans; also investigating reinforcement learning for adaptive interaction policies

Current Status

• System fully implemented• Early version “tested” on Pearl at Longwood Elder

Care Facility in Oakmont, PA• Later version currently being tested on handhelds,

without sensing/ with simple (RFID-based sensing), with TBI patients from U of M Med Rehab Clinic

• Larger scale wireless sensing technology being developed and integrated into Autominder in the lab, for field testing later this year

Key Challenges for Cognitive Orthotics

• Technological– Advanced AI Techniques

– HCI

– Sensor Networks for Inference of Daily Activities

– Mechanisms to Ensure Privacy and Security

• Policy– Mechanisms to Ensure Privacy

– Reimbursement Policies

For More Information…

www.eecs.umich.edu/~pollackm

Extra Slides Follow….

The Plan Manager

• Maintains up-to-date record of client’s planned activities and their execution status– Eating– Hydrating – Toileting– Medicine-taking – Exercise – Social activities – Doctors’ appointments– etc.

How Does it Work?

• Models constraints on future actions– Lunch takes between 25 and 35 minutes – Take meds within one hour of finishing lunch – Watch the news at either 6pm or at 11pm

• Performs efficient constraint processing when key events occur:– New planned activity added.– Existing activity modified or deleted.– Planned activity performed.– Critical time bounds passed.

Small Example

ClientPlan

1. New Activity2. Mod/Deletion3. Activity Execution4. Passed Time Bound

PLAN MANAGER

:0 MS – LE :60“Take meds within 1 hour of lunch”

LE = 12:15“Lunch ended at 12:15”-----------------------------12:15 MS 13:15“Take meds by 1:15”

Client Client ModelerModeler

Plan Plan ManagerManager

IntelligentIntelligentReminderReminderGeneratorGenerator

ClientPlan

Activity Info

Inferred Activity

Sensor Data

Reminders

Client Model Info

Activity Info

Preferences

Plan Updates

ClientModel

Autominder Architecture

CM: Client Modeler

Given what can be observed• Sensor input: client moved to kitchen • Clock time: at 7:23 a.m.• Client plan: breakfast should be eaten between 7 and 8• Model of previous actions: client has not yet eaten breakfast• Learned patterns: 82% of the time, client starts breakfast between 7:10 and 7:25• Reminder information: we issued a reminder at 7:21

Infers what has been done• Client Activity: probability that client has begun breakfast

How Does it Work?

• Models probabilistic relations among observations and actions

• Performs Bayesian update, extended to handle temporal relations• Asks for confirmation when needed!

started

breakfast

breakfastreminder issued

went tokitchen

reminder kitchen start-breakfast Y Y .95 Y N .10 N Y .8 N N .03

Client Client ModelerModeler

Plan Plan ManagerManager

IntelligentIntelligentReminderReminderGeneratorGenerator

ClientPlan

Activity Info

Inferred Activity

Sensor Data

Reminders

Client Model Info

Activity Info

Preferences

Plan Updates

ClientModel

Autominder Architecture

Intelligent Reminders

• Decides whether and when to issue reminders• Given a client’s plan and its execution status:

– Easy to generate reminders• Remind at earliest possible time of each action

– Harder to “remind well”• Maximize likelihood of appropriate performance of

ADLs and other key activities• Facilitate efficient performance• Avoid annoying client• Avoid making client overly reliant

How Does it Work (Now)?

LB D

TV

Midnight

8:00 16:0012:00

12:00

LB D

TV

Midnight

8:00 16:0012:00

12:00

LB D

TV

Midnight

8:00 16:0012:00

12:00

8:30 12:32

How Will it Work?

• Use reinforcement learning to deduce an optimal reminding strategy

• Model the system as a Markov decision process that– Senses the environment

– Decides what action to perform

– Receives a “payoff”

and then “learn” the best policy after repeated interactions

Current Status of Autominder

• V.0 (Autominder + Pearl) field-tested for client acceptability on Pearl at Longwood Elderly Care Facility in Oakmont, PA, summer, 2001

• V.1 of Autominder implemented – Java, Lisp on Wintel machines

• Data collection with three Oakmont residents completed summer 2002; with Ann Arbor TBI patient summer 2003

• Systematic field-testing to begin momentarily with TBI patients

• Many projects going on to develop technology to support (older) individuals with cognitive impairment

• With the potential to have a huge impact • But still lots of issues to resolve:

– A host of scientific questions and engineering challenges

• Sensor interpretation

• Interface design

• . . .

– Question of cost and reimbursement structure

– Privacy, privacy, privacy!

Conclusions

Acknowledgements

Autominder• DTP/PM:

– Ioannis Tsamardinos– Sailesh Ramakrishnan – Cheryl Orosz

• CM:– Dirk Colbry– Bart Peintner

• IRG:– Colleen McCarthy– Matt Rudary

• System Integration:– Laura Brown– Martina Gierke– Peter Schwartz– Joe Taylor

Funders•National Science Foundation•Intel Corporation[Supporting Technology: DARPA, AFOSR]

PearlSebastian Thrun, Mike Montemerlo, Joelle Pineau, Nick Roy

Rest of the Nursebot TeamJacqueline Dunbar-Jacob, Sandra Engberg, Judy Matthews, Sara Keisler, Don Chiarulli, Jennifer Goetz