Chapter 5
AI to Scale and Improve Access to Healthcare Services
In developing countries, inequities between urban and rural health services is a serious problem. A shortage of qualified healthcare providers is a major cause of the unavailability and low quality of healthcare in rural areas. Some studies have shown that the application of computer-assisted or AI-based medical techniques could improve healthcare outcomes in rural areas and in developing countries.
Challenge
Pneumonia is the number one infectious cause of death for children under age five worldwide. According to UNICEF, pneumonia claimed the lives of more than 880,000 children in 2016—most of whom were less than two years old. Treatment for pneumonia is not the primary issue as readily available antibiotics are highly successful. Misdiagnosis is the main challenge, especially in the areas where access to healthcare is very limited.
Solutions
Brian Turyabagye and two colleagues from Makerere University in Kampala, Olivia Koburongo and Besufekad Shifferaw, founded Mama-Ope in 2016 to develop an AI-based approach for diagnosing pneumonia in children.
- They designed a wearable medical device: a smart jacket that has five microphones effectively working as wearable stethoscopes to measure lung sounds from multiple locations on a child’s torso.
- The Mama-Ope team programmed a signal processing algorithm to provide the best diagnostic insight available from the audio recordings. The goal: Determine when the distinctive crackle sound of pneumonia was recorded. Heuristically, the distinctive lung sounds include wheezing and crackling.
- The team and an expert from MathWorks explored the signals in MATLAB using signal processing and wavelet techniques. They found distinctive features that were present throughout the signal.
- They isolated these distinctive features to train a machine learning algorithm in MATLAB that can predict the cases where pneumonia is present.
An AI-based approach to diagnosing pneumonia in children. (Image credit: RAEng/Brett Eloff)
Here is what pneumonia sounds like. (Audio Source: thesimtech.com)
Results
The jacket is designed to be used at remote clinics and schools. Even locations without medical personnel or a computer can use the jacket for a speedy diagnosis. The jacket connects to a mobile phone app via Bluetooth® and records and analyzes the collected data. It then sends the results to a health care professional who can make an informed diagnosis without requiring an in-person examination of the child. UNICEF has already expressed interest in helping Mama-Ope bring its technology to schools, hospitals, and clinics in the hard-hit pneumonia areas of sub-Saharan Africa such as Uganda, Kenya, Tanzania, Ethiopia, and Nigeria.
Challenge
Cervical cancer is curable when caught early, but early diagnosis depends on routine screening. In rural areas, particularly in low- and middle-income countries, few people can access this care. Improving access to screening is critical, because people in these areas account for 85% of cervical cancer cases.
Cervisense delivers on-the-spot cervical cancer screening results with machine learning. (Image credit: Akshita Sachdeva)
Solution
Akshita Sachdeva and Bonny Dave started a company called Satin Healthtech in India and developed a product called Cervisense. Cervisense comprises a small camera mounted on a speculum and a tablet-based program that delivers on-the-spot cervical cancer screening results. It’s an optical imaging–based screening tool that helps automate and enhance the accuracy of cervical cancer screening. The program applies machine learning techniques on the images, which is taken by a healthcare worker during patient’s cervix exam and provides them with a preliminary diagnosis score to guide their care recommendations.
Results
Satin Healthtech has completed its Cervisense prototype, and they have interest from gynecologists and oncologists in India. A few months after Satin Healthtech was founded, the World Health Organization announced the Global Strategy for Cervical Cancer Elimination calling for higher screening rates.
Challenge
An estimated half a billion people worldwide have a mental illness, and the cost in the United States alone reaches $500 billion a year. Many people benefit from psychological counseling, but gaps remain. Mental health assessment is highly subjective, and the focus on prevention is limited. Additionally, diagnoses are often missed, and there’s little real-time intervention.
Solution
Improving mental health with continuous patient monitoring has the potential to help patients suffering from mental health issues such as anxiety and depression. Sentio’s Feel program is a wristband and phone app that tracks users’ emotional states, offers regular mental and physical exercises, and connects them to a therapist once a week.
- The system uses machine learning and signal processing algorithms developed in MATLAB to identify patterns in data collected from the wearable device, such as combinations of biomarkers that indicate different emotional states.
- Sentio first started with psychology literature describing which physical signals are most indicative of which emotions. Then they fine-tuned the models by asking people wearing the emotion sensor to describe their feelings. The models also classified the readings, and if the emotion labels differed from what the user described, the models updated themselves to work better next time.
- These algorithms are deployed on cloud-based Amazon® Web Services (AWS) servers that monitor the emotional state of the patient and feed results back to the app.
Machine learning and signal processing use physiological sensor data to detect emotions.
Machine learning and signal processing use physiological sensor data to detect emotions.
Results
The system has been tested with hundreds of users. The system adapts itself to each individual user, and people have reported that they feel like the Feel Emotion Sensor is really getting to know them.
