1

MLivin’ it

Engineering productive health, anytime, anywhere

Goh Chen Gang, Low Zi An, Marion Pang, Xie Mingrou

ASN Hackathon 2021

Problem: Non Communicable Diseases

2

https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases

Images from BioRender

>70%

annually

3

Smoking

​

Physical inaction

​

Poor Diet

Drinking

NON COMMUNICABLE DISEASES (NCDs)

Risk Factors

Can be changed!

https://www.who.int/health-topics/noncommunicable-diseases#tab=tab_3

https://www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases

Images from BioRender

​

Case Study: Singapore

Mean Salt/day = 9 g >>> 5 g (Maximum)

Sugar/day = 60 g >>> 25 g (WHO)

Calories/day = 2470 >>> 2200/1800 (HPB)

4

National Nutrition Survey (2018)

(Males) (Females)

https://www.straitstimes.com/singapore/health/salt-and-sugar-intake-way-too-high-though-fruit-and-veg-consumption-improving-hpb

WHO: World Health Organisation

HPB: Health Promotion Board

COVID-19 - Challenges

5

Wellbeing

The effects of social isolation on well-being and life satisfaction during pandemic | Humanities and Social Sciences Communications (nature.com)

2020-12.pdf (essex.ac.uk)

https://www.forbes.com/sites/eladnatanson/2020/10/22/are-business-productivity-apps-ruling-our-lives/?sh=6aaa98482a03

Images from BioRender

Productivity

Apps → Crucial facilitator

Current Apps

Health

Productivity

Wellbeing

6

Noom

Lose it!

Todoist

Any.do

Headspace

Happify

https://www.healthline.com/health/diet-and-weight-loss/top-iphone-android-apps#lose-it

https://www.tomsguide.com/round-up/best-productivity-apps

https://www.menshealth.com/health/g22842908/best-health-and-fitness-apps/

https://play.google.com/store/apps/details?id=com.todoist

https://play.google.com/store/apps/details?id=com.anydo

https://play.google.com/store/apps/details?id=com.getsomeheadspace.android&hl=en_US

https://play.google.com/store/apps/details?id=com.happify.happifyinc&hl=en_SG&gl=US

​

Not integrative!

Our Idea - App for Productive Health

7

Wearable device

8

CALORIES INTAKE

CALORIES BURNT

GENDER

BMI

SLEEP ANALYSIS

TIME

Wearable device

Food recognition (mobile phone)

SMART SCORE +

SUGGESTIONS

Workout

Eat/Snack

>

<

Personal Health & Wellbeing

User input data

9

PRESET DAY MODE

TO-DO LIST

DAY RATINGS

SCREEN TIME

SMART ANALYSIS

Productivity

WORK

CHILL

SOCIALS

STAY-AT-HOME

OUTDOORS

CATEGORISED APPS

• SOCIAL
• COMMUNICATIONS
• ENRICHMENT
• ENTERTAINMENT
• SHOPPING
• CREATIVITY
• FITNESS

REMINDER

POSTPONE

PROCRASTINATION SCORE

MOOD

SATISFACTION

PRODUCTIVITY SCORE

WEARABLE DEVICE DATA

Key Advantages

10

1. Real time, longitudinal
2. Integrative
3. Ex vivo, non invasive
4. Context aware
5. Multimodal Data
6. Personalised Data
7. Dynamic
8. Memory

ML Algorithms

11

Computer Vision

Time-series analytics

Image Recognition

Deep Convolutional Neural Network

Transfer Learning

Time-series Decomposition

Trend/Seasonality Analysis

Computer Vision for food recognition

12

Images analyzed on server

Nutritional values pulled from online database

Computer Vision for food recognition

13

Our quick implementation: �~90% ACCURACY

Training images pulled from Google

DCNN ResNet-18

Transfer Learning

DCNN: Deep Convolutional Neural Network

Computer Vision for food recognition

14

Training with 5 classes, n=150

Computer Vision for food recognition

15

Successful identification:

Unsuccessful identification:

App Demo Video

16

Path to successful deployment

17

Launch the app on the app store (freemium model)

Market analysis of relevant stakeholders

Test the app on a select group of volunteers and debug

Integrate with nutrition datasets

​

Do alpha and beta testing

Continue with support for users and continual updates

Thanks!

Any questions ?

​

18