SlideShare a Scribd company logo
1 of 47
Download to read offline
Assessing the Quality of
Doctor Consultations using ML
Pranjal Aswani
Data Engineer @ Halodoc
Who are we?
PATIENT
S
PAYERS PROVIDERS
PATIENT
S
PAYERS PROVIDERS
Halodoc upgrades
relationship between Patients,
Payers and Providers
Halodoc services
ONLINE OFFLINE
But how are online consultations helping
people?
Number of Physicians
Per 10,000 Population
OECD 2018
30
UK
28.1
Number of Physicians
Per 10,000 Population
OECD 2018
30
UK
28.1
US
24.5
South Korea
22.0
Singapore
19.5
World
average
13.9
Number of Physicians
Per 10,000 Population
OECD 2018
30 0
UK
28.1
US
24.5
South Korea
22.0
Singapore
19.5
World
average
13.9
Malaysia
12.0
Thailand
3.9
Indonesia
3.0
Number of Physicians
Per 10,000 Population
OECD 2018
Healthcare unavailability
because of archipelagic
geography
Halodoc online
consultations
vs
User’s journey for online
consultation and providing
feedback
Consultation user journey
Feedback user journey
Why assess consultation feedbacks?
Consultation feedbacks
50% of the total consultations
are rated
As of
December’ 19
1 Bad
experience
with a doctor
Negative
app rating
Customer feedbacks help in providing
actionable feedback to the doctors
How do we assess the quality of
consultations?
- Manually look at random consultations by different doctors, which contain
feedback and point out the mistakes the doctors are making
- Quantitative: Look at metrics a consultation produces (num messages sent,
length, notes, etc) and flag obvious consultations (0 messages, late response from
doctor, etc)
- Qualitative: Look at how the consultation did on SAPE
- Check if in the end, the patient’s problem was solved or not
Taking help of humans to analyse
consultations
● SAPE is a modified version of SOAP, which is a global standard for assessing consultations
○ Subjective: retrieve information from patient by asking questions
■ Main symptom (high fever), Additional Symptom (body ache)
○ Objective: vitals, measurements (temperature, BP)
○ Assessment: explain the ailment to the patient and why it happened
■ Differential diagnosis (viral fever different from flu), Possible etiology (viral)
○ Planning: necessary steps for the patient to get better and preventative measures
■ Lifestyle modification (rest, light food), recommendation (paracetamol)
○ Etiquette: politeness and empathy towards the patient (replacement for O)
■ Opening (hello) and closing Etiquette (good bye!)
● O is difficult for online consultation and we measure E, instead
Consultation metrics based on “SAPE”
How can we use Machine Learning for
this?
Consultation quality with SAPE
USER COMMUNICATES
THEIR PROBLEM
FOLLOW UP QUESTIONS
FROM OUR DOCTOR
Subjective
DIAGNOSIS BY DOCTOR
MEDICINE
RECOMMENDATIONS
Assessment
Planning
Consultation quality with SAPE … contd
DOCTOR ETIQUETTESEtiquette
Consultation quality with SAPE … contd
● Given an anonymized chat consultation transcript between the doctor & patient,
automate SAPE scoring
● SAPE scores are in the range {0, 0.5, 1.0}
● Goals
○ Actionable feedback to the doctors
○ Improve the consultations quality on the platform
○ Auto summarisation of chat consultations ( => Doctor notes)
Problem statement
● NLP in bahasa Indonesia
○ Limited NLP resources
○ NLP with medical terms
○ Translate to English?
● Avoiding bias in the dataset
○ Positive and negative feedback consultations
○ Consultation category
● Training dataset
○ Equally distributed labelled data
○ Oversampling or Undersampling
Tech Challenges
Evaluated Approaches
- Quantitative (numbers)
- Qualitative (context)
Numeric features based on RED (Responsiveness, Effort and Diligence) scores created
to assess doctors
- Responsiveness: Acceptance Time , First reply time, messages with response time
more than 1 min
- Effort: Doctor-patient message ratio
- Diligence: Notes depth, completed time
Created features based on this domain knowledge
Quantitative-Round 1: using downvoted consultations data
- Average response time
- Average length of message
- Number of messages sent
- Duration of consultation
- Chat closed by
- eRx issued or not
- Doctor patient chat ratio
- Number of questions asked (based on a question classifier)
What kind of data did we start with?
● Team of intern doctors manually labelling the data
● Around 6K labelled dataset
○ Data in google sheets
○ RED scores at a doctor level + SAPE scores at a consultation level
● Bias in the data (only thumbs down consultations were considered)
Find relations between quantitative features (RED scores) + SAPE scores (tags)
- Collate all the available data, clean it and remove duplicates
- Each sub-category is given a 0.5 or 0.0 score (available tags)
- Combine sub-category scores go get scores for the category itself (0, 0.5, 1.0)
- Train decision trees and neural networks using quantitative features
- Hyperparameter (variables of the models which can be messed with for a better
output) tuning
Test accuracies
Production accuracies
A coin toss would’ve given us a better
accuracy score than this
Why didn’t it work?
- Less data (around 7K consultations after cleaning the data, all downvoted)
- Question classifier (a basic dictionary approach) was not that accurate (~40%)
- We had more data for certain scores (1.0) and less for others (0.0, 0.5)
(imbalanced dataset)
How do you make it better?
- Collect better data
- Feature engineering to find more possible features (such as relation of age of patients to the
consultation duration)
- Sampling
- Create a better question classifier (needed for Subjective category in SAPE)
Data Collection
- Clean and better structured data
- Input for RED scores
- Sentence level tagging
- No PII information exposed to
the intern doctors, as compared
to earlier
- Generating reports from the
data for doctors
- Average of ~300 consultations
being tagged per day (as
compared to 100-150 per day)
Quantitative-Round 2: using better data + sampling + better features
- Sampling (over and under, per category) to fix imbalanced dataset
- Created a better question classifier using Support Vector Machine + TfIdf ( 43% more accurate )
- Feature engineering to find more meaningful features
- Age of patient/doctor
- Gender of patient/doctor
- Type of consultation (general, pediatric, OBGYN)
- Re-generate features and fill missing values with averages
- More data
- More enthusiasm!!
Subjective category results on test set
Previous results on test set
Our models failed because the
distribution of our features
across the 3 scores was
almost identical
The models couldn’t find any
patterns because there weren’t
any!
But why didn’t you do it
earlier?
→ Because we only had one
type of data: downvoted
Rethink the
features!
Qualitative: finding context in a consultation
What do we have?
- Sentences tagged at a category level per consultation
- Scores for each of the categories for a consultation (0, 0.5, 1 across S, A and P)
Main idea:
- For each of the categories, there will be words and word pairs (n-grams) which occur only in
sentences of that category
- Exploit this for each of the categories
- To predict the score for Assessment (0 or 1) based on the chat sentences available
- Total sentences: ~40 K
- Take all A sentences of a consultation and combine them into 1 sentence
- Generate Tf-Idf vector for all those consultation sentences
- Train an SVM to predict the value of A for that consultation as 0 or 1
- ~70% accuracy on test set
Pilot: Assessment classifier
Category classifiers
- Create individual category classifiers
to classify sentences as S, A or P
sentences
- Use the classified sentences per
category and feed it to the sub
category classifiers
- Train sub-category (main symptom,
additional symptom for Subjective
category) classifiers to predict scores
0, 0.5 using sentences for that
category
Results?
Category classifier test accuracy
Sub category classifier test accuracy
The final models were chosen
after experimenting with a
dozen different kinds
algorithms
Prod results
Subjective accuracy of 70% (+25% improvement over quantitative techniques)
Assessment accuracy of ~62% (~45% improvement on accuracy)
Planning accuracy of ~57% (~40% improvement over previous algorithms).
● Quantity and Quality of the dataset
● Avoid bias in dataset
● Metrics to measure the impact
● Setting expectations with the business stakeholders
● Working with uncertainty
Learnings
Next steps
- Get more data
- Tag patient level sentences to get the full context of the consultation
- Create better models using word2vec
- Repeat
QUESTIONS?
Excited?
Join us!
careers.india@halodoc.com
www.halodoc.com
blogs.halodoc.io
● https://www.linkedin.com/in/pranjalaswani/
● https://www.linkedin.com/in/rdurgam/
THANK
YOU

More Related Content

What's hot

Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...
Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...
Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...CORE Group
 
Jisc learning analytics MASHEIN Jan 2017
Jisc learning analytics MASHEIN Jan 2017Jisc learning analytics MASHEIN Jan 2017
Jisc learning analytics MASHEIN Jan 2017Paul Bailey
 
Jisc learninganalytics dec2016
Jisc learninganalytics dec2016Jisc learninganalytics dec2016
Jisc learninganalytics dec2016Paul Bailey
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentationnirvdrum
 
Brief Introduction to the 12 Steps of Evaluation Data Cleaning
Brief Introduction to the 12 Steps of Evaluation Data CleaningBrief Introduction to the 12 Steps of Evaluation Data Cleaning
Brief Introduction to the 12 Steps of Evaluation Data CleaningJennifer Morrow
 
[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation
[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation
[WI 2017] Affective Prediction By Collaborative Chains In Movie RecommendationYONG ZHENG
 
Quantitative Method Assignment Help
Quantitative Method Assignment HelpQuantitative Method Assignment Help
Quantitative Method Assignment HelpStat Analytica
 
Melissa Informatics - Data Quality and AI
Melissa Informatics - Data Quality and AIMelissa Informatics - Data Quality and AI
Melissa Informatics - Data Quality and AImelissadata
 
Data analysis – using computers for presentation
Data analysis – using computers for presentationData analysis – using computers for presentation
Data analysis – using computers for presentationNoonapau
 
Strategies for Practical Active Learning, Robert Munro
Strategies for Practical Active Learning, Robert MunroStrategies for Practical Active Learning, Robert Munro
Strategies for Practical Active Learning, Robert MunroRobert Munro
 
EAHIL 2017 Building Better Metrics
EAHIL 2017 Building Better MetricsEAHIL 2017 Building Better Metrics
EAHIL 2017 Building Better MetricsAlan Fricker
 
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System Quality
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System QualityOlivier Blais. Model Validation Tips and Tricks to Ensure AI System Quality
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System QualityLviv Startup Club
 

What's hot (20)

Chatbots: Automated Conversational Model using Machine Learning
Chatbots: Automated Conversational Model using Machine LearningChatbots: Automated Conversational Model using Machine Learning
Chatbots: Automated Conversational Model using Machine Learning
 
Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...
Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...
Operations Research: Methods, Challenges, Emerging Lessons, and Opportunities...
 
Jisc learning analytics MASHEIN Jan 2017
Jisc learning analytics MASHEIN Jan 2017Jisc learning analytics MASHEIN Jan 2017
Jisc learning analytics MASHEIN Jan 2017
 
Jisc learninganalytics dec2016
Jisc learninganalytics dec2016Jisc learninganalytics dec2016
Jisc learninganalytics dec2016
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentation
 
Machine Learning For Stock Broking
Machine Learning For Stock BrokingMachine Learning For Stock Broking
Machine Learning For Stock Broking
 
Image Analytics for Retail
Image Analytics for RetailImage Analytics for Retail
Image Analytics for Retail
 
Blended act
Blended actBlended act
Blended act
 
Brief Introduction to the 12 Steps of Evaluation Data Cleaning
Brief Introduction to the 12 Steps of Evaluation Data CleaningBrief Introduction to the 12 Steps of Evaluation Data Cleaning
Brief Introduction to the 12 Steps of Evaluation Data Cleaning
 
Analytics in Online Retail
Analytics in Online RetailAnalytics in Online Retail
Analytics in Online Retail
 
[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation
[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation
[WI 2017] Affective Prediction By Collaborative Chains In Movie Recommendation
 
Machine Learning in ICU mortality prediction
Machine Learning in ICU mortality predictionMachine Learning in ICU mortality prediction
Machine Learning in ICU mortality prediction
 
Quantitative Method Assignment Help
Quantitative Method Assignment HelpQuantitative Method Assignment Help
Quantitative Method Assignment Help
 
Melissa Informatics - Data Quality and AI
Melissa Informatics - Data Quality and AIMelissa Informatics - Data Quality and AI
Melissa Informatics - Data Quality and AI
 
Data analysis – using computers for presentation
Data analysis – using computers for presentationData analysis – using computers for presentation
Data analysis – using computers for presentation
 
Strategies for Practical Active Learning, Robert Munro
Strategies for Practical Active Learning, Robert MunroStrategies for Practical Active Learning, Robert Munro
Strategies for Practical Active Learning, Robert Munro
 
EAHIL 2017 Building Better Metrics
EAHIL 2017 Building Better MetricsEAHIL 2017 Building Better Metrics
EAHIL 2017 Building Better Metrics
 
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System Quality
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System QualityOlivier Blais. Model Validation Tips and Tricks to Ensure AI System Quality
Olivier Blais. Model Validation Tips and Tricks to Ensure AI System Quality
 
Analytics for offline retail: Using Advanced Machine Learning
Analytics for offline retail: Using Advanced Machine LearningAnalytics for offline retail: Using Advanced Machine Learning
Analytics for offline retail: Using Advanced Machine Learning
 
A3 thinking nhsiq 2014
A3 thinking  nhsiq 2014A3 thinking  nhsiq 2014
A3 thinking nhsiq 2014
 

Similar to Assessing Doctor Consultations using ML

Ppr r101-session-3-smart
Ppr r101-session-3-smartPpr r101-session-3-smart
Ppr r101-session-3-smartSharmiBhat
 
TTIPEC: Monitoring and Evaluation (Session 2)
TTIPEC: Monitoring and Evaluation (Session 2)TTIPEC: Monitoring and Evaluation (Session 2)
TTIPEC: Monitoring and Evaluation (Session 2)Research to Action
 
ACT Implementation in SAFTINet
ACT Implementation in SAFTINetACT Implementation in SAFTINet
ACT Implementation in SAFTINetMarion Sills
 
Implementing Point-of-Care PROMs
Implementing Point-of-Care PROMsImplementing Point-of-Care PROMs
Implementing Point-of-Care PROMsMarion Sills
 
1. What is the problem2. Why is the problem important and r
1. What is the problem2. Why is the problem important and r1. What is the problem2. Why is the problem important and r
1. What is the problem2. Why is the problem important and rTatianaMajor22
 
LMuldoon JRayner - 2015 CACHC Conference Presentation
LMuldoon JRayner - 2015 CACHC Conference PresentationLMuldoon JRayner - 2015 CACHC Conference Presentation
LMuldoon JRayner - 2015 CACHC Conference Presentationcachc
 
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...U.S. News Healthcare of Tomorrow
 
UXPA 2021: How do you know your users feel satisfied
UXPA 2021: How do you know your users feel satisfied   UXPA 2021: How do you know your users feel satisfied
UXPA 2021: How do you know your users feel satisfied UXPA International
 
Recommending Scientific Papers: Investigating the User Curriculum
Recommending Scientific Papers: Investigating the User CurriculumRecommending Scientific Papers: Investigating the User Curriculum
Recommending Scientific Papers: Investigating the User CurriculumJonathas Magalhães
 
EAP Research Findings: Relevance for Today's EAPs
EAP Research Findings: Relevance for Today's  EAPsEAP Research Findings: Relevance for Today's  EAPs
EAP Research Findings: Relevance for Today's EAPsBernie McCann
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxstirlingvwriters
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxstirlingvwriters
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxstirlingvwriters
 
Remarks from the professor on milestone 1 and for milestone 2A.docx
Remarks from the professor on milestone 1 and for milestone 2A.docxRemarks from the professor on milestone 1 and for milestone 2A.docx
Remarks from the professor on milestone 1 and for milestone 2A.docxsodhi3
 
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...Asma Ben Abacha
 
Right Things Right: ZEF Evaluation
Right Things Right: ZEF EvaluationRight Things Right: ZEF Evaluation
Right Things Right: ZEF EvaluationJordan Epp
 
Quality Assessment and Quality Improvement in Support Services for Persons wi...
Quality Assessment and Quality Improvement in Support Services for Persons wi...Quality Assessment and Quality Improvement in Support Services for Persons wi...
Quality Assessment and Quality Improvement in Support Services for Persons wi...Buntinx Training & Consultancy
 
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....CORE Group
 

Similar to Assessing Doctor Consultations using ML (20)

Ppr r101-session-3-smart
Ppr r101-session-3-smartPpr r101-session-3-smart
Ppr r101-session-3-smart
 
TTIPEC: Monitoring and Evaluation (Session 2)
TTIPEC: Monitoring and Evaluation (Session 2)TTIPEC: Monitoring and Evaluation (Session 2)
TTIPEC: Monitoring and Evaluation (Session 2)
 
ACT Implementation in SAFTINet
ACT Implementation in SAFTINetACT Implementation in SAFTINet
ACT Implementation in SAFTINet
 
7INNOVA Lesson 5 slides for T26
7INNOVA Lesson 5 slides for T267INNOVA Lesson 5 slides for T26
7INNOVA Lesson 5 slides for T26
 
Implementing Point-of-Care PROMs
Implementing Point-of-Care PROMsImplementing Point-of-Care PROMs
Implementing Point-of-Care PROMs
 
1. What is the problem2. Why is the problem important and r
1. What is the problem2. Why is the problem important and r1. What is the problem2. Why is the problem important and r
1. What is the problem2. Why is the problem important and r
 
Data Analysis
Data AnalysisData Analysis
Data Analysis
 
LMuldoon JRayner - 2015 CACHC Conference Presentation
LMuldoon JRayner - 2015 CACHC Conference PresentationLMuldoon JRayner - 2015 CACHC Conference Presentation
LMuldoon JRayner - 2015 CACHC Conference Presentation
 
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...
Using Hospital Data Insights to Improve Clinical Quality at UCLA Medical Cent...
 
UXPA 2021: How do you know your users feel satisfied
UXPA 2021: How do you know your users feel satisfied   UXPA 2021: How do you know your users feel satisfied
UXPA 2021: How do you know your users feel satisfied
 
Recommending Scientific Papers: Investigating the User Curriculum
Recommending Scientific Papers: Investigating the User CurriculumRecommending Scientific Papers: Investigating the User Curriculum
Recommending Scientific Papers: Investigating the User Curriculum
 
EAP Research Findings: Relevance for Today's EAPs
EAP Research Findings: Relevance for Today's  EAPsEAP Research Findings: Relevance for Today's  EAPs
EAP Research Findings: Relevance for Today's EAPs
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docx
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docx
 
Nursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docxNursing Quality Key Indicator.docx
Nursing Quality Key Indicator.docx
 
Remarks from the professor on milestone 1 and for milestone 2A.docx
Remarks from the professor on milestone 1 and for milestone 2A.docxRemarks from the professor on milestone 1 and for milestone 2A.docx
Remarks from the professor on milestone 1 and for milestone 2A.docx
 
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...
NLP Methods for Medical Question Answering (Philips AI 2020) | Dr. Asma Ben A...
 
Right Things Right: ZEF Evaluation
Right Things Right: ZEF EvaluationRight Things Right: ZEF Evaluation
Right Things Right: ZEF Evaluation
 
Quality Assessment and Quality Improvement in Support Services for Persons wi...
Quality Assessment and Quality Improvement in Support Services for Persons wi...Quality Assessment and Quality Improvement in Support Services for Persons wi...
Quality Assessment and Quality Improvement in Support Services for Persons wi...
 
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....
Seven steps for Use Routine Information to Improve HIV/AIDS Program_Snyder_5....
 

More from GDG Cloud Bengaluru

Making cloud native deployments easy with Buildpack
Making cloud native deployments easy with BuildpackMaking cloud native deployments easy with Buildpack
Making cloud native deployments easy with BuildpackGDG Cloud Bengaluru
 
Cloud spanner architecture and use cases
Cloud spanner architecture and use casesCloud spanner architecture and use cases
Cloud spanner architecture and use casesGDG Cloud Bengaluru
 
What's new with serverless on google cloud
What's new with serverless on google cloud What's new with serverless on google cloud
What's new with serverless on google cloud GDG Cloud Bengaluru
 
Managing serverless workloads with knative
Managing serverless workloads with knativeManaging serverless workloads with knative
Managing serverless workloads with knativeGDG Cloud Bengaluru
 
Building observable infrastructure and code
Building observable infrastructure and codeBuilding observable infrastructure and code
Building observable infrastructure and codeGDG Cloud Bengaluru
 
Kubernetes best practices with GKE
Kubernetes best practices with GKEKubernetes best practices with GKE
Kubernetes best practices with GKEGDG Cloud Bengaluru
 
Anthos Application Modernization Platform
Anthos Application Modernization PlatformAnthos Application Modernization Platform
Anthos Application Modernization PlatformGDG Cloud Bengaluru
 
Intro to GKE and app deployment with Kubernetes
Intro to GKE and app deployment with KubernetesIntro to GKE and app deployment with Kubernetes
Intro to GKE and app deployment with KubernetesGDG Cloud Bengaluru
 
Google Cloud - Stand Out Features
Google Cloud - Stand Out FeaturesGoogle Cloud - Stand Out Features
Google Cloud - Stand Out FeaturesGDG Cloud Bengaluru
 

More from GDG Cloud Bengaluru (11)

Making cloud native deployments easy with Buildpack
Making cloud native deployments easy with BuildpackMaking cloud native deployments easy with Buildpack
Making cloud native deployments easy with Buildpack
 
Cloud spanner architecture and use cases
Cloud spanner architecture and use casesCloud spanner architecture and use cases
Cloud spanner architecture and use cases
 
What's new with serverless on google cloud
What's new with serverless on google cloud What's new with serverless on google cloud
What's new with serverless on google cloud
 
Managing serverless workloads with knative
Managing serverless workloads with knativeManaging serverless workloads with knative
Managing serverless workloads with knative
 
Introduction to OpenFaas
Introduction to OpenFaasIntroduction to OpenFaas
Introduction to OpenFaas
 
Building observable infrastructure and code
Building observable infrastructure and codeBuilding observable infrastructure and code
Building observable infrastructure and code
 
Kubernetes best practices with GKE
Kubernetes best practices with GKEKubernetes best practices with GKE
Kubernetes best practices with GKE
 
Serverless solutions on GCF
Serverless solutions on GCFServerless solutions on GCF
Serverless solutions on GCF
 
Anthos Application Modernization Platform
Anthos Application Modernization PlatformAnthos Application Modernization Platform
Anthos Application Modernization Platform
 
Intro to GKE and app deployment with Kubernetes
Intro to GKE and app deployment with KubernetesIntro to GKE and app deployment with Kubernetes
Intro to GKE and app deployment with Kubernetes
 
Google Cloud - Stand Out Features
Google Cloud - Stand Out FeaturesGoogle Cloud - Stand Out Features
Google Cloud - Stand Out Features
 

Recently uploaded

IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsEnterprise Knowledge
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking MenDelhi Call girls
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel Araújo
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking MenDelhi Call girls
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfsudhanshuwaghmare1
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptxHampshireHUG
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsJoaquim Jorge
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonAnna Loughnan Colquhoun
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountPuma Security, LLC
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessPixlogix Infotech
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking MenDelhi Call girls
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?Igalia
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
Real Time Object Detection Using Open CV
Real Time Object Detection Using Open CVReal Time Object Detection Using Open CV
Real Time Object Detection Using Open CVKhem
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 

Recently uploaded (20)

IAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI SolutionsIAC 2024 - IA Fast Track to Search Focused AI Solutions
IAC 2024 - IA Fast Track to Search Focused AI Solutions
 
08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men08448380779 Call Girls In Friends Colony Women Seeking Men
08448380779 Call Girls In Friends Colony Women Seeking Men
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
04-2024-HHUG-Sales-and-Marketing-Alignment.pptx
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path Mount
 
Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your Business
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Real Time Object Detection Using Open CV
Real Time Object Detection Using Open CVReal Time Object Detection Using Open CV
Real Time Object Detection Using Open CV
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 

Assessing Doctor Consultations using ML

  • 1. Assessing the Quality of Doctor Consultations using ML Pranjal Aswani Data Engineer @ Halodoc
  • 4. PATIENT S PAYERS PROVIDERS Halodoc upgrades relationship between Patients, Payers and Providers
  • 6. But how are online consultations helping people?
  • 7. Number of Physicians Per 10,000 Population OECD 2018
  • 8. 30 UK 28.1 Number of Physicians Per 10,000 Population OECD 2018
  • 11. Healthcare unavailability because of archipelagic geography Halodoc online consultations vs
  • 12. User’s journey for online consultation and providing feedback
  • 16. Consultation feedbacks 50% of the total consultations are rated As of December’ 19
  • 17. 1 Bad experience with a doctor Negative app rating
  • 18. Customer feedbacks help in providing actionable feedback to the doctors
  • 19. How do we assess the quality of consultations?
  • 20. - Manually look at random consultations by different doctors, which contain feedback and point out the mistakes the doctors are making - Quantitative: Look at metrics a consultation produces (num messages sent, length, notes, etc) and flag obvious consultations (0 messages, late response from doctor, etc) - Qualitative: Look at how the consultation did on SAPE - Check if in the end, the patient’s problem was solved or not Taking help of humans to analyse consultations
  • 21. ● SAPE is a modified version of SOAP, which is a global standard for assessing consultations ○ Subjective: retrieve information from patient by asking questions ■ Main symptom (high fever), Additional Symptom (body ache) ○ Objective: vitals, measurements (temperature, BP) ○ Assessment: explain the ailment to the patient and why it happened ■ Differential diagnosis (viral fever different from flu), Possible etiology (viral) ○ Planning: necessary steps for the patient to get better and preventative measures ■ Lifestyle modification (rest, light food), recommendation (paracetamol) ○ Etiquette: politeness and empathy towards the patient (replacement for O) ■ Opening (hello) and closing Etiquette (good bye!) ● O is difficult for online consultation and we measure E, instead Consultation metrics based on “SAPE”
  • 22. How can we use Machine Learning for this?
  • 23. Consultation quality with SAPE USER COMMUNICATES THEIR PROBLEM FOLLOW UP QUESTIONS FROM OUR DOCTOR Subjective
  • 26. ● Given an anonymized chat consultation transcript between the doctor & patient, automate SAPE scoring ● SAPE scores are in the range {0, 0.5, 1.0} ● Goals ○ Actionable feedback to the doctors ○ Improve the consultations quality on the platform ○ Auto summarisation of chat consultations ( => Doctor notes) Problem statement
  • 27. ● NLP in bahasa Indonesia ○ Limited NLP resources ○ NLP with medical terms ○ Translate to English? ● Avoiding bias in the dataset ○ Positive and negative feedback consultations ○ Consultation category ● Training dataset ○ Equally distributed labelled data ○ Oversampling or Undersampling Tech Challenges
  • 28. Evaluated Approaches - Quantitative (numbers) - Qualitative (context)
  • 29. Numeric features based on RED (Responsiveness, Effort and Diligence) scores created to assess doctors - Responsiveness: Acceptance Time , First reply time, messages with response time more than 1 min - Effort: Doctor-patient message ratio - Diligence: Notes depth, completed time Created features based on this domain knowledge Quantitative-Round 1: using downvoted consultations data - Average response time - Average length of message - Number of messages sent - Duration of consultation - Chat closed by - eRx issued or not - Doctor patient chat ratio - Number of questions asked (based on a question classifier)
  • 30. What kind of data did we start with? ● Team of intern doctors manually labelling the data ● Around 6K labelled dataset ○ Data in google sheets ○ RED scores at a doctor level + SAPE scores at a consultation level ● Bias in the data (only thumbs down consultations were considered)
  • 31. Find relations between quantitative features (RED scores) + SAPE scores (tags) - Collate all the available data, clean it and remove duplicates - Each sub-category is given a 0.5 or 0.0 score (available tags) - Combine sub-category scores go get scores for the category itself (0, 0.5, 1.0) - Train decision trees and neural networks using quantitative features - Hyperparameter (variables of the models which can be messed with for a better output) tuning
  • 32. Test accuracies Production accuracies A coin toss would’ve given us a better accuracy score than this
  • 33. Why didn’t it work? - Less data (around 7K consultations after cleaning the data, all downvoted) - Question classifier (a basic dictionary approach) was not that accurate (~40%) - We had more data for certain scores (1.0) and less for others (0.0, 0.5) (imbalanced dataset) How do you make it better? - Collect better data - Feature engineering to find more possible features (such as relation of age of patients to the consultation duration) - Sampling - Create a better question classifier (needed for Subjective category in SAPE)
  • 34. Data Collection - Clean and better structured data - Input for RED scores - Sentence level tagging - No PII information exposed to the intern doctors, as compared to earlier - Generating reports from the data for doctors - Average of ~300 consultations being tagged per day (as compared to 100-150 per day)
  • 35. Quantitative-Round 2: using better data + sampling + better features - Sampling (over and under, per category) to fix imbalanced dataset - Created a better question classifier using Support Vector Machine + TfIdf ( 43% more accurate ) - Feature engineering to find more meaningful features - Age of patient/doctor - Gender of patient/doctor - Type of consultation (general, pediatric, OBGYN) - Re-generate features and fill missing values with averages - More data - More enthusiasm!!
  • 36. Subjective category results on test set Previous results on test set
  • 37. Our models failed because the distribution of our features across the 3 scores was almost identical The models couldn’t find any patterns because there weren’t any! But why didn’t you do it earlier? → Because we only had one type of data: downvoted Rethink the features!
  • 38. Qualitative: finding context in a consultation What do we have? - Sentences tagged at a category level per consultation - Scores for each of the categories for a consultation (0, 0.5, 1 across S, A and P) Main idea: - For each of the categories, there will be words and word pairs (n-grams) which occur only in sentences of that category - Exploit this for each of the categories
  • 39. - To predict the score for Assessment (0 or 1) based on the chat sentences available - Total sentences: ~40 K - Take all A sentences of a consultation and combine them into 1 sentence - Generate Tf-Idf vector for all those consultation sentences - Train an SVM to predict the value of A for that consultation as 0 or 1 - ~70% accuracy on test set Pilot: Assessment classifier
  • 40. Category classifiers - Create individual category classifiers to classify sentences as S, A or P sentences - Use the classified sentences per category and feed it to the sub category classifiers - Train sub-category (main symptom, additional symptom for Subjective category) classifiers to predict scores 0, 0.5 using sentences for that category
  • 41. Results? Category classifier test accuracy Sub category classifier test accuracy The final models were chosen after experimenting with a dozen different kinds algorithms
  • 42. Prod results Subjective accuracy of 70% (+25% improvement over quantitative techniques) Assessment accuracy of ~62% (~45% improvement on accuracy) Planning accuracy of ~57% (~40% improvement over previous algorithms).
  • 43. ● Quantity and Quality of the dataset ● Avoid bias in dataset ● Metrics to measure the impact ● Setting expectations with the business stakeholders ● Working with uncertainty Learnings
  • 44. Next steps - Get more data - Tag patient level sentences to get the full context of the consultation - Create better models using word2vec - Repeat