The document discusses using AWS technologies to enable ambient intelligence and assisted living solutions for connected home environments. It describes using sensors to collect data about inhabitants' activities, vital signs, and environment. AWS IoT Analytics is used to ingest sensor data, enrich it using pipelines, and analyze it to monitor safety, detect anomalies, and provide alerts. Challenges include privacy, sensor accuracy, and data handling. Best practices involve remote management, technology choices based on home type, privacy-preserving configurations, and non-intrusive designs.
3. Related breakouts
IOT208 - What's new with AWS IoT analytics services?
GPSTEC413 - Best practices and design patterns for building IoT
solutions on AWS
ARC339-R1 - [REPEAT 1] Best practices for IoT architecture using AWS
smart product solution
API305-R - [REPEAT] Building serverless machine-learning workflows
4. Agenda
What is Ambient Intelligence (AmI)?
Connected Homes for Assisted Living
AWS Technologies for Assisted Living
AWS IoT Analytics Deep Dive
Demo
Challenges and Best Practices
7. ambient
adjective
am·bi·ent| ˈam-bē-ənt
existing or present on all sides :
encompassing
Ambient intelligence brings ‘smarts’ to a connected home, which
means a home able to:
• assure more security to its dwellers
• communicate with the outside world
• save energy
• be more comfortable
What is Ambient Intelligence (AmI)?
8. Use Case – Ambient Assisted Living
Need for AmI in Assisted living
• Aging population
• Mobility detection
• Vital statistics monitoring
• Location detection
• Activity monitoring
• Eliminating social isolation
10. Connected Homes for Assisted Living - Building Blocks
Sensors
Depth Sensors, Motion Sensors, Thermal Sensors
Event Orchestration and Visualization
Conversational AI, Actuator
Intelligent Event Recognition
Computer Vision Algorithms, Anomaly Detection
11. Assisted Living – Technology Enablers
• Advancement in home automation (or Domotics) technologies
• Wireless Sensor Networks
• Wearable Biomedical Sensors and Activity Trackers
• Environment Sensors
• Protocol Interoperability Hubs – ZigBee/ZigBee PRO, BLE, LoRaWAN
• IoT Gateway Platforms
• Networking Integration with Care Centers & family members
• Edge Computing and ML Advancements
• Predictive modeling and Anomaly detection
12. 360° View of Ambient Assisted Living
• Detection of fall
or sudden injury
• Communicate
with family
members
• Track physical
regimen
• Detect Leading
Indicators
• Co-relate medical
diagnosis
Vital
Parameter
Checks
Activity
Tracker
User
Location
Detection
Extended
Assistance
14. AWS Technologies for Assisted Living
• Amazon FreeRTOS
• AWS IoT Core and AWS IoT Greengrass
• Amazon Sagemaker Neo
• AWS IoT Analytics
• Amazon QuickSight
• AWS IoT Events
• Alexa for Business
15. AWS IoT Analytics for Assisted Living
Connected
Elderly Home
CameraThermostat
Medical
House
Door lock
Elderly
Person
Wearables
IoT Core
IoT
Analytics
Channel
IoT
Analytics
Pipeline
Medical
Institution
Medical History Enrichment Weather Information
Weather
Station
Pipeline Enrichment Activities
Inhabitant
safety
alerts
IoT
Analytics
Datastore
Monitoring
Dashboards
Extended Assistance
Medical
Support
Family
Members
Facilities
Management
Machine Learning Inference
Pre-built Machine
Learning Models
23. Challenges
• Privacy
• Sensor Accuracy
• Data Store and Compute Capacity
Best Practices
• Remote Management and Monitoring
• New Houses vs Old Houses (BLE vs ZigBee)
• Privacy preserving configurations
• Non-Intrusive
Welcome, my name is Neel Sendas – Sr Technical Account Manager at AWS and my collegue Krishnan G
Before we get started let’s give you a quick reminder of other related breakout sessions –
IoT-208 – What’s new in IoT Analytics
GPSTEC413 – Partner focused talk on best practices around IoT solutions from the field
ARC339- Another solution that highlights best practice
API305 – How you use multiple AI/ML services to build a serverless ML solution
So here’s the agenda, we are gonna talk about What is AmI intelligence?
How it applies to a particular use case – Ambient Intelligence in Assisted Living home
Talk about a few AWS solutions that one can use to build an AmI Solution
Walk you through a one of the AWS solutions in particular – AWS IoT Analytics that one can use for AmI
And then we are gonna see a demo
So what is Ambient Intelligence?
You can read the definitation here but essesially it’s an intelligent environment that fullfils the needs of individuals in that environment
It’s an environment that adapts to your needs and takes intelligence decisions on your behalf
It’s an environment that orchestrates data collected from different sensors and smart devices and creates context awareness for the environment you are in
It weaves data collected from all the sensors around and creates a pattern
And as a results provides better and security lifestyle for people in though environments
It takes the Human Computer Interaction to a different level – It takes the sensor data gathered from all the smart devices and sensors around us and apply the intelligence of bigdata to machine learning to generate insights that would not have to been otherwise possible to generate.
Once particular area where Ambient Intelligence can make a lot of difference in assisted living for our aging polutation
You might talk that Aging population in developed countries is a becoming a big problem. More than 20% of the population in 3 countries Germany, Italy, and Japan are over 65 years. There are over 68,000 people in Japan are centenarians.
We need the necessary trained staffs and infrastructure to take care of this problem. That’s when AmI comes in. They can multiply the efforts of a lean staff and maintain quality of care
Additionally - to err is human. Decision making is difficult. Sensors can reduce or eliminate errors
IoT sensors and devices make it easier to manage this problem
Mobility devices – sensors such a accelerometer, gyroscope can monitor monitor movement, fall, detect, abnormal movements,
Assisted living homes in eurpo are using smart tiles to monitor pressure and activity
Collect vitals – BP moinitors, ECG
Location detection – using RFID and GPS sensors can locate whether an individual and stayed outside in the backyard (especially important for alzeimers patients)
Elimimate social isolation by using Intelligent conversational tools
Aging population – By 2034
Error rate, challengesMobility detection - Fall, Immobility, slow movementVital statistics monitoring - Respiratory rate, BP, temperatureLocation detection - Front door, backyard, living roomEating habits, fluid intake, medication frequencyEleminating social isolation
Healthcare is a great concern
diag
Here are the key building blocks of Ambient Intelligence in Connected homes
Sensors – depth sensors can help regonize people without compromising privacy
Thermal sensors – heat sensors for human activity
Intelligent event recognition – using computer vision algorithm, outlier events, compute at the edge
Identifying and managing dense activities - Multiple synchronous and asynchronous activites.
Real time synchronization tools.
Not sure how Reinforcement Learning & Temporal Learning happen in a connected home ?
Krishnan to improve upon this slide
We can improve upon this slide
Non-intrusive
Security and privacy – healthcare is a very intimate space. We need to respect privacy
Interoperability and lack of standardization means you have to do the heavy lifting of protocol interoperability
IoT Analytics, Lambda on the edge using IoT Analytics can help.
Imbalanced data set – Oversampling, SMOTE, Sythenic dataset, and transfer learning
Sensor accuracy is an issue