Explainable deep learning models for healthcare - CDSS 3

Explainable deep learning models for healthcare - CDSS 3

This course is part of Informed Clinical Decision Making using Deep Learning Specialization

Instructor: Fani Deligianni

Sponsored by Louisiana Workforce Commission

1,673 already enrolled

4 modules

Gain insight into a topic and learn the fundamentals.

4.6

(15 reviews)

Intermediate level

Recommended experience

30 hours to complete

3 weeks at 10 hours a week

Flexible schedule

Learn at your own pace

4 modules

Gain insight into a topic and learn the fundamentals.

4.6

(15 reviews)

Intermediate level

Recommended experience

30 hours to complete

3 weeks at 10 hours a week

Flexible schedule

Learn at your own pace

What you'll learn

Program global explainability methods in time-series classification
Program local explainability methods for deep learning such as CAM and GRAD-CAM
Understand axiomatic attributions for deep learning networks
Incorporate attention in Recurrent Neural Networks and visualise the attention weights

Details to know

Shareable certificate

Add to your LinkedIn profile

Assessments

5 assignments

Taught in English

See how employees at top companies are mastering in-demand skills

Learn more about Coursera for Business

Build your subject-matter expertise

This course is part of the Informed Clinical Decision Making using Deep Learning Specialization

When you enroll in this course, you'll also be enrolled in this Specialization.

Learn new concepts from industry experts
Gain a foundational understanding of a subject or tool
Develop job-relevant skills with hands-on projects
Earn a shareable career certificate

Earn a career certificate

Add this credential to your LinkedIn profile, resume, or CV

Share it on social media and in your performance review

There are 4 modules in this course

This course will introduce the concepts of interpretability and explainability in machine learning applications. The learner will understand the difference between global, local, model-agnostic and model-specific explanations. State-of-the-art explainability methods such as Permutation Feature Importance (PFI), Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanation (SHAP) are explained and applied in time-series classification. Subsequently, model-specific explanations such as Class-Activation Mapping (CAM) and Gradient-Weighted CAM are explained and implemented. The learners will understand axiomatic attributions and why they are important. Finally, attention mechanisms are going to be incorporated after Recurrent Layers and the attention weights will be visualised to produce local explanations of the model.

Deep learning models are complex and it is difficult to understand their decisions. Explainability methods aim to shed light to the deep learning decisions and enhance trust, avoid mistakes and ensure ethical use of AI. Explanations can be categorised as global, local, model-agnostic and model-specific. Permutation feature importance is a global, model agnostic explainabillity method that provide information with relation to which input variables are more related to the output.

What's included

6 videos8 readings1 assignment1 discussion prompt5 ungraded labs

6 videosTotal 71 minutes

Welcome video - Explainable Deep Learning Models for Healthcare0 minutesPreview module
Interpretability vs Explainability16 minutes
'Explainability' in Healthcare Applications17 minutes
Taxonomy of Explainability Methods17 minutes
Model Agnostic Explainability Methods10 minutes
Permutation Feature Importance in Time Series Data8 minutes

8 readingsTotal 390 minutes

The importance of explainable prediction models in healthcare10 minutes
Explainable Artificial Intelligence - Taxonomy60 minutes
Model Agnostic Explainability30 minutes
Permutation Feature Importance60 minutes
Practical Exercise: Interpretability of the MLP model using Permutation Feature Importanceg60 minutes
Practical Exercise: Interpretability of the CNN model using Permutation Feature Importance60 minutes
Practical Exercise: Interpretability of the LSTM model using Permutation Feature Importance60 minutes
Explainability models in ECG50 minutes

1 assignmentTotal 30 minutes

End of week 1 quiz30 minutes

1 discussion promptTotal 10 minutes

Week 1 - Your experience10 minutes

5 ungraded labsTotal 50 minutes

Permutation feature importance for classifying heart beats using a CNN10 minutes
Light - Permutation feature importance for classifying heart beats using a CNN10 minutes
Permutation feature importance for classifying heart beats using an LSTM10 minutes
Light - Permutation feature importance for classifying heart beats using an LSTM10 minutes
Permutation feature importance for classifying heart beats using a multi-layer perceptron10 minutes

Local explainability methods provide explanations on how the model reach a specific decision. LIME approximates the model locally with a simpler, interpretable model. SHAP expands on this and it is also designed to address multi-collinearity of the input features. Both LIME and SHAP are local, model-agnostic explanations. On the other hand, CAM is a class-discriminative visualisation techniques, specifically designed to provide local explanations in deep neural networks.

What's included

5 videos7 readings1 assignment1 discussion prompt7 ungraded labs

5 videosTotal 47 minutes

Local Interpretable Model Agnostic Explanations (LIME)9 minutesPreview module
LIME in Time-Series Classification9 minutes
Shapley Additive Explanations11 minutes
Model-Specific Explanations: Visualisation Methods12 minutes
CAM in Time-Series Classification4 minutes

7 readingsTotal 320 minutes

Why Should I Trust You?30 minutes
Practical Exercise: Interpretability of heartbeat classification using LIME and an NNMLP model60 minutes
Practical Exercise: Interpretability of heartbeat classification using LIME and a CNN model60 minutes
Practical Exercise: Interpretability of heartbeat classification using LIME and an LSTM model60 minutes
A Unified Approach to Interpreting Model Predictions40 minutes
Practical Exercise: Interpretability of CNN models using Class Activation Maps10 minutes
Class Activation Mapping60 minutes

1 assignmentTotal 30 minutes

End of week 2 quiz30 minutes

1 discussion promptTotal 10 minutes

Week 2 - Your experience10 minutes

7 ungraded labsTotal 65 minutes

Interpretability of heartbeat classification using a CNN model and Class Activation Maps10 minutes
Interpretability of heartbeat classification using a CNN model and CAM10 minutes
LIME interpretability for heartbeat classification with a convolutional neural network10 minutes
Interpretability of heartbeat classification using an LSTM model and CAM10 minutes
LIME interpretability for heartbeat classification with a long short-term memory network10 minutes
Light - LIME interpretability for heartbeat classification with a long short-term memory network5 minutes
LIME interpretability for heartbeat classification with a multi-layer perceptron10 minutes

GRAD-CAM is an extension of CAM, which aims to a broader application of the architecture in deep neural networks. Although, it is one of the most popular methods in explaining deep neural network decisions, it violates key axiomatic properties, such as sensitivity and completeness. Integrated gradients is an axiomatic attribution method that aims to cover this gap.

What's included

4 videos6 readings1 assignment1 discussion prompt7 ungraded labs

4 videosTotal 36 minutes

Gradient Weighted Class Activation Maps11 minutesPreview module
Grad-CAM in Time-Series Classification8 minutes
Integrated Gradients12 minutes
Integrated Gradients in Time Series Classification3 minutes

6 readingsTotal 310 minutes

GRAD - Class Activation Mapping 30 minutes
Practical Exercise: Interpretability of the CNN model using Gradient-weighted Class Activation Mapping60 minutes
Practical Exercise: Interpretability of the LSTM model using Gradient-weighted Class Activation Mapping60 minutes
Axiomatic Attribution for Deep Networks40 minutes
Practical Exercise: Interpretability of the CNN model using Integrated Gradients60 minutes
Practical Exercise: Interpretability of the LSTM model using Integrated Gradients60 minutes

1 assignmentTotal 30 minutes

End of week 3 quiz30 minutes

1 discussion promptTotal 10 minutes

Week 3 - Your experience10 minutes

7 ungraded labsTotal 70 minutes

Interpretability of heartbeat classification using a CNN model and Grad-CAM10 minutes
Interpretability of heartbeat classification using integrated gradients and a CNN model10 minutes
Light - Interpretability of heartbeat classification using integrated gradients and a CNN model10 minutes
Interpretability of heartbeat classification using an LSTM model and Grad-CAM10 minutes
Light - Interpretability of heartbeat classification using an LSTM model and Grad-CAM10 minutes
Interpretability of heartbeat classification using integrated gradients and an LSTM model10 minutes
Light - Interpretability of heartbeat classification using integrated gradients and an LSTM model10 minutes

Attention in deep neural networks mimics human attention that allocates computational resources to a small range of sensory input in order to process specific information with limited processing power. In this week, we discuss how to incorporate attention in Recurrent Neural Networks and autoencoders. Furthermore, we visualise attention weights in order to provide a form of inherent explanation for the decision making process.

What's included

3 videos3 readings2 assignments1 discussion prompt4 ungraded labs

3 videosTotal 34 minutes

Attention in Deep Learning13 minutesPreview module
Taxonomy of Attention12 minutes
Attention and Explainability8 minutes

3 readingsTotal 190 minutes

Survey on Attention Mechanisms10 minutes
Practical Exercise: Classification of heartbeats using an LSTM with attention mechanism90 minutes
Practical Exercise: Interpretability of the LSTM model with attention mechanism90 minutes

2 assignmentsTotal 60 minutes

End of course summative quiz30 minutes
End of week 4 quiz30 minutes

1 discussion promptTotal 10 minutes

Week 4 - Your experience10 minutes

4 ungraded labsTotal 30 minutes

Interpretability of heartbeat classification using an LSTM model with attention mechanism10 minutes
Light - Interpretability of heartbeat classification using an LSTM model with attention mechanism5 minutes
Heartbeat classification using an LSTM model with attention mechanism10 minutes
Light - Heartbeat classification using an LSTM model with attention mechanism5 minutes