Made at Metis: Deep Learning to Detect Pneumonia & Predicting Spotify Track Skips

By Carlos Russo • August 31, 2020

This post features two projects from recent graduates of our data science bootcamp. Take a look at what's possible to create in just 12 weeks. 


Pneumonia Detection: Pushing the Boundaries of Human Ability with Deep Learning
Jenny Wang, Strategy and Operations Analyst at Google

Getting pneumonia may be more common than you think. According to bootcamp graduate Jenny Wang, it's the most common reason for U.S. children to be hospitalized and it's the most common cause of hospital admissions for U.S. adults other than women giving birth. For her final project, she chose to build a deep learning model to detect pneumonia from chest X-Ray images. In a blog post summarizing the project, she writes: 

"It turns out that chest X-rays have long been considered the best tool to detect any form of pneumonia. However, studies have shown that even experienced radiologists often have a hard time correctly identifying whether something is an infiltrate (a substance denser than air, such as pus or fluid) on the X-Ray. This causes delays in diagnosis, which increases the disease severity and associated mortality rate. If humans aren’t able to correctly diagnose the disease early, maybe…deep learning can help!"

In her post, she goes into detail about her approach to the project, the data collection, the analysis, image processing, modeling, and more. It's a great read, and you can check it out here.

Predicting Spotify Track Skips
Austin Poor, Metis Bootcamp Graduate

For his third of five projects during the bootcamp, graduate Austin Poor worked on what he calls "a slightly simplified version of the Spotify Sequential Skip Prediction Challenge." As he details in a blog post about the project, Spotify supplied two main sets of information for this competition.

The first consisted of data about user listening sessions, including details like track ID, position in the session, if the track was skipped or not, and more. The second dataset consisted of information about the tracks themselves, including duration, popularity in the U.S., and release year. 

In his post, read how he took a subset of this data and used feature engineering and more to get results and plans for future work. Read it here.

    _____

    See more examples of Metis student projects here


    Similar Posts

    data science
    Made at Metis: An Album Discovery Tool & A Voting Recommendation Engine

    By Carlos Russo • September 25, 2020

    This post features two projects from recent graduates of our data science bootcamp. Take a look at what's possible to create in just 12 weeks.

    data science
    How to Become a Data Scientist

    By Carlos Russo • April 16, 2021

    Data science jobs are plentiful in today’s job market. Read on to learn about what data scientists do, what kinds of data science-related jobs are available, and how to become a data scientist.

    data science
    Made at Metis: Predicting Stock Performance & AI-Generated Guided Meditations

    By Emily Wilson • July 27, 2020

    This post features two projects from recent graduates of our data science bootcamp. Take a look at what's possible to create in just 12 weeks.