Careers in AI— AI in Healthcare

This article is part of the ‘Careers in AI’ series. In this article we explore how Artificial Intelligence and Healthcare interconnect and reveal what to expect if you decide to pursue a career in this intersection.

Authors: Andre Frade & Jenny Shim

The application of AI to the healthcare sector is in its early stages and is quickly revolutionising the way we prevent, diagnose and treat diseases.

What is the healthcare sector about?

The Healthcare sector refers to the economy segment that is dedicated to the management of health. The sector comprises all businesses and institutions that play a central role in the research, diagnosis, treatment, nursing and management of illness, diseases and injuries.

The healthcare sector is one of the world’s largest and fast-growing sectors, shaping economies and driving research worldwide.

Why consider machine learning in healthcare?

Healthcare data is abundant and complex. AI techniques have the ability to augment the human cognitive capabilities and provide unprecedented insights into the analysis and interpretation of such data. So much so, that AI based healthcare solutions have already shown to outperform medical experts.

The ability to process information with an increased speed, precision and accuracy helps save precious time, decrease risks, and increase success rate when delivering medical solutions to patients. Furthermore, AI is also very suited to the optimisation of resource allocation, which helps to increase patient turnover and decrease medical costs, making healthcare more accessible to all.

Future

AI technologies have already proven their potential to transform how health and care are accessed, maintained and restored. AI is expected to continue providing more and better solutions in the healthcare sector. We mainly expect to see advances in the way new drugs are discovered, developed and brought to market. Likewise, there are high hopes on personalised care to increase the success rate of treatments, specially to complex neurological diseases, mental health conditions, and cancer.

Types of Applications

Some of the AI technologies of high importance to healthcare include machine learning, deep learning, natural language processing, rule based expert systems, robotics and robotic automation processes. These serve the basis to most of the AI solutions that have been developed and applied to practices such as preventive medicine, patient monitoring, diagnosis processes, treatment optimisation, drug discovery and personalised medicine.

Virtual Nurses
Virtual nurses are always available to provide support to patients. These online assistants interact via text or voice and can perform a range of tasks, that includes answering general questions, providing quick diagnostics, guiding patients to the correct care, or even closely monitoring patients. Babylon and Buoy Health are examples of companies developing solutions like these.

Diagnosis
AI healthcare tools are extraordinary at uncovering patterns from data. Technologies like this can help doctors to integrate patient data and perform holistic analysis to enable faster and more accurate clinical judgement. The ability to analyse large amounts of data — health condition, lifestyle, genetic, etc — has shown a great potential to uncover predisposition or the early identification of a particular disease or condition.

AI powered vision tools are the step up to AI assisted diagnosis. Image analysis is expected to decongest the healthcare system or even make healthcare more accessible to remote areas. These tools are built to receive pictures of patient conditions and provide professional medical advice. PathAI and Poscia are examples of companies developing this type of solution.

Personalised Treatments
The personalisation of healthcare treatments is quickly evolving. Many companies are dedicated to the development of methods that can adjust medical treatments to the profile of each patient. These methods are often powered by AI tools that can thoroughly analyse vast amounts of personal data and derive customised solutions, that are expected to have a faster response, higher success rate and lead to less side effects. Benevolent AI and Tempus are putting great effort towards this type of solution.

Surgical Robots
Surgical robots are used to augment the experience and skill of doctors to help them perform complex procedures with increased precision and control. Robot assisted surgeries are considered minimally invasive and often lead to less complications, less pain and a quicker recovery time after the operation. Hospitals have also started to notice the impact of the reduction in costs and patient’s hospital stay. Vicarious Surgical and Auris Health are examples of companies operating in this space.

System Optimisation
Healthcare and medical technologies produce huge amounts of patient data every day. AI tools have been developed to process this data faster and more efficiently, helping institutions to manage and optimise patient flow and resource allocation. Oventus is an example of software developed for this purpose.

Drug Research
The traditional journey of a drug from the research lab to the patient is long and costly: an average of 12 years and US $360 million per approved drug, according to the California Biomedical Research Association. Research and industrial organisations are quickly noticing the insight that AI technologies can provide in this space. Drug research is one of the most recent applications for AI in healthcare and has already shown its potential to transform drug development pipelines, cutting both the time to market for new drugs and their costs. BioXcel Therapeutics and Atomwise are examples of companies dedicated to AI assisted drug research.

Types of companies

The part of the AI sector dedicated to Healthcare is one of the world’s fastest growing within the AI field. Most businesses are small to medium size companies, and often have a big research component and prototyping focus. From finding new links between genetic codes to driving surgery assisting robots, artificial intelligence is driving the development of modern healthcare. Some companies operating in the space include:

· Pharmaceuticals — SaliencyAI, Owkin, Benevolent AI
· Medical research — Google Health, Oncora Medical, Verily
· Medical Diagnostics — IBM Watson Health, Corti
· Medical Imaging — Poscia

Types of Jobs

Researcher/engineer in the following areas:
· Radiology entails the development of AI tools to interpret imaging results and help clinicians detect details easily missed by the human eye. These mostly employ deep learning technology to detect a range of diseases and disorders.
· Screening entails the development of AI systems to describe and evaluate the outcome of surgeries such as maxillo-facial surgery and cleft palate therapy. These tools are also developed to detect skin, breast and prostate cancer. Deep learning and computer vision are popular requirements in this area.
· Drug interactions, where scientists use natural language processing to identify drug interactions in medical literature or develop AI powered methods to identify novel potential interactions and their effects.

NLP Specialist
Work with customers to develop Natural Language Processing based solutions that can handle medical written data, mostly dedicated to the automation of administrative tasks or to medical literature screening.

Robotic Surgery
Work on the development and successful integration of new surgical robots into hospitals. Projects often aim for the creation/expansion of robotic surgery software and hardware.

Example of interview questions

Technical
· Solve unusual and unique problems using Bayesian inference
· Write code to handle unusual or unlikely situations
· Describe algorithmic differentiation/solve the questions
· Solve problems in subjects such as information theory, logistic regression, calculus, automatic differentiation, deep learning, classification etc
· Conduct exploratory data analysis, machine learning, data mining and statistics to build solutions with actionable insights from high-volume behaviour data.
· Define and measure success KPIs/metrics through applied data science that help improvements in conversion or customer satisfaction.

Non-Technical
· Talk about coding projects that you have done in the past
· Talk about your background and how much understanding you have in advanced statistical techniques to solve real problems
· Talk about a time when you demonstrated excellent communication and presentation skills, ability to explain complex data
· Do you have strong organisation and prioritisation skills?

Links & References

https://builtin.com/artificial-intelligence/artificial-intelligence-healthcare
https://en.wikipedia.org/wiki/Artificial_intelligence_in_healthcare
https://towardsdatascience.com/microsofts-data-scientist-interview-questions-4c6f3a62ef64
https://en.wikipedia.org/wiki/Artificial_intelligence_in_healthcare#:~:text=Artificial%20intelligence%20in%20healthcare%20is,complicated%20medical%20and%20healthcare%20data.
https://www.forbes.com/sites/bernardmarr/2018/07/27/how-is-ai-used-in-healthcare-5-powerful-real-world-examples-that-show-the-latest-advances/?sh=83895005dfbe
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6616181/
https://builtin.com/artificial-intelligence/artificial-intelligence-healthcare
https://chem.libretexts.org/Courses/Intercollegiate_Courses/Cheminformatics_OLCC_(2019)
https://www.predictiveanalyticstoday.com/what-is-healthcare-industry/