AI is transforming the practice of medicine. It’s helping doctors diagnose patients more accurately, make predictions about patients’ future health, and recommend better treatments. This Specialization will give you practical experience in applying machine learning to concrete problems in medicine.
Class Deals by MOOC List - Click here and see Coursera's Active Discounts, Deals, and Promo Codes.
Machine learning is a powerful tool for prognosis, a branch of medicine that specializes in predicting the future health of patients. In this second course, you’ll walk through multiple examples of prognostic tasks. You’ll then use decision trees to model non-linear relationships, which are commonly observed in medical data, and apply them to predicting mortality rates more accurately. Finally, you’ll learn how to handle missing data, a key real-world challenge.
These courses go beyond the foundations of deep learning to teach you the nuances in applying AI to medical use cases. This course focuses on tree-based machine learning, so a foundation in deep learning is not required for this course. However, a foundation in deep learning is highly recommended for course 1 and 3 of this specialization. You can gain a foundation in deep learning by taking the Deep Learning Specialization offered by deeplearning.ai and taught by Andrew Ng.
Course 2 of 3 in the AI for Medicine Specialization.
What You Will Learn
- Walk through examples of prognostic tasks
- Apply tree-based models to estimate patient survival rates
- Navigate practical challenges in medicine like missing data
Syllabus
WEEK 1
Linear prognostic models
Build a linear prognostic model using logistic regression, then evaluate the model by calculating the concordance index. Finally, improve the model by adding feature interactions.
WEEK 2
Prognosis with Tree-based models
Tune decision tree and random forest models to predict the risk of a disease. Evaluate the model performance using the c-index. Identify missing data and how it may alter the data distribution, then use imputation to fill in missing data, in order to improve model performance.
WEEK 3
Survival Models and Time
This week, you will work with data where the time that a disease occurs is a variable. Instead of predicting just the 10-year risk of a disease, you will build more flexible models that can predict the 5 year, 7 year, or 10 year risk.
WEEK 4
Build a risk model using linear and tree-based models
This week, you will fit a linear model, and a tree-based risk model on survival data, to customize a risk score for each patient, based on their health profile. The risk score represents the patient’s relative risk of getting a particular disease. You will then evaluate each model’s performance by implementing and using a concordance index that incorporates time to event and censored data.
