Deep Learning in Healthcare Part 2: Future & Predictions

View Part 1 of the discussion here.

The application of artificial intelligence and deep learning in healthcare and medicine is often quoted to grow tenfold in 5 years, from algorithms that learn to recognise complex patterns within rich medical data, to analysing real world evidence for personalised medicine, to discovering the sequence specificities of DNA-binding proteins and how it can aid genome diagnostics.

As part of our ongoing speaker Q&A series, we asked deep learning and healthcare experts for their predictions for the field over the next 5 years, the risks involved with AI integration, areas for disruption and more.

Diogo Moitinho de Almeida is a data scientist, software engineer, and hacker. He joined Enlitic as Senior Data Scientist in 2014, where he develops deep learning algorithms for medical diagnosis.

Sobia Hamid holds a PhD in Epigenetics from the University of Cambridge, and is Founder of Data Insights Cambridge, a nonprofit community of over 800 data science practitioners.

Olexandr Isayev is a Research Scientist at UNC Eshelman School of Pharmacy, University of North Carolina at Chapel Hill, where his research focuses on making sense of chemical data with molecular modelling and machine learning.

Vinay Kumar is a researcher in nanotech and co-Founder and CEO of, a developer platform for artificial intelligence with deep learning tools in language, vision, text, speech, dialogues and reasoning.

Michael Nova is Chief Innovation Officer and a co-Founder of Pathway Genomics, as well as the inventor of the Pathway-IBM Watson deep machine learning and AI mobile application Panorama/OME.

Naveen Rao is a co-Founder and CEO of Nervana Systems. Trained as both a computer architect and neuroscientist, he founded Nervana to use biological inspiration and take computation in new directions.

What developments can we expect to see in deep learning in the next 5 years?
Diogo: I believe we can expect to see spatial attention that works well enough that it will be considered a prerequisite to any state of the art visual model. This will greatly reduce computational requirements and enable analysis of higher dimensional problems that are fairly important in medicine. Today, deep learning models are used for only a subset of machine learning problems. I believe that we will see that subset grow significantly as we improve our abilities to optimize deep models and integrate the best parts of the "shallow" algorithms as differentiable components into deep ones. Finally, the data requirements for successful deep learning applications will be significantly lower: they will need less labeled data, they will effectively utilize noisy and incomplete data, and they will pull together data from multiple modalities to find patterns that we know should exist, but as of today have no ability to find.

Naveen: “Recent advances in deep learning technology will help us solve many of the world’s most complex challenges,” said Steve Jurvetson, Partner at DFJ. “By developing deep learning solutions that are faster, easier and less expensive to use, Nervana is democratizing deep learning and fueling advances in medical diagnostics, image and speech recognition, genomics, agriculture, finance, and eventually across all industries."

Broad application of deep learning into all aspects of our lives will happen in the next 5 years. The way we access healthcare, shop, farm, or interact with other people will all be shaped by learning machines. Deep learning will allow us to better and more efficiently use resources and drive down the cost of services. In addition, our experiences with machines will be personalized as websites and devices adapt to our individual preferences.

On the research side, unsupervised learning will advance in the next 5 years. About 90% of all data in the world is unlabeled. This means there is no description of the meaning the data or what inferences can be drawn (images, sounds, GPS tracking data, exercise data, etc.). Unsupervised learning is the next big frontier for finding useful inferences in data.

Vinay: Today we can see a lot of single function neural networks but going forward we will see more and more multi-functional neural nets that can perform multiple tasks and can remember the decision flows needed to perform complex tasks. This can transform the scope and scale of multi-functional robots and can truly bring us to a point where we will witness mass production of robots.

Michael: Should expect to see much deeper models, models that can learn from many fewer training cases compared to today's models, and substantial advances in unsupervised learning. We should expect to see even more accurate and useful speech and visual recognition systems. Expect deep learning methods to be applied to increasingly multi-modal problems with more structure in the data. Deep learning algorithms will become so efficient that they will be able to run on cheap mobile devices, even without extra hardware support or prohibitive memory overhead.

Olexandr: As patients within 5 years, we will see results of the deep learning revolution in practice. It takes certain time to translate success in research into clinical practice.

1. DL will revolutionize diagnostic healthcare. Doctors are overwhelmed with diagnostic data: MRIs, CTs, X-rays, biopsy, etc. The convolutional neural nets (CNN) will extremely efficiently process this wealth of image and volumetric (3D) data. CNN will automatically analyze and segment image, find a suspected disease and provide objective outcome with proper confidence scores.
2. Computer and AI system will prevent many medical errors. We humans can make errors. Doctors are working long hours, under stress and other real life circumstances. AI assistants can help making right decision or prevent a potential error.
3. Digital Pharmacist. This AI system will take a prescription from a patient and instantly dispense a drug, suggest a generic.
4. Inexpensive AI systems will provide universal healthcare to millions and millions of people across the globe, especially in poor countries of Africa and Latin America.

Overall, I see appearance of extremely smart AI assistants [think like J.A.R.V.I.S. (in Iron Man movie) for MDs]. They will be interactive real time expert advisors; doctors will talk with them directly by means of natural language.

What area of deep learning advancements excites you most?
Sobia: Deep learning technologies for medical imaging offer incredible advancements in improved resolution, breadth and speed of analysis and diagnosis. There are also sophisticated advancements being made across computational biology research, for instance in the application of deep neural networks to predicting protein structures. The application of deep learning methodologies to clinical trial data is also showing great potential for drug development and wider therapeutic interventions.

Diogo: All of deep learning is really exciting to me, but the software side of things excites me the most because it just makes everything better. The tools that exist today are far from perfect, but even as is they've enabled the massive amount of results that has given the field its reputation. I believe this is just the beginning, and that we are, in the very near future, going to make software that will allow us to dwarf what has been done so far and make the tools we use today look like they're from the deep learning stone age.

Naveen: Personally, I think healthcare is an important area of advancement. The cost of care is skyrocketing, and machine learning can simultaneously bring down the price of care as well as improve the quality. Using machines to aid with diagnostics means that the same high level of competency can be easily applied to more people.

Vinay: 'AI system building' process is the most exciting segment. So far, we have been seeing humans building machines and guiding them through rules (code) while implementing in the process. This, human intervention, is a big bottleneck for 'AI' Adoption'. Even though everyone wants to use AI, not everyone can build a reliable system. However, if this expertise is given to a machine, then a machine can start building another machine. These AI systems with experiences can become better with time and experience and will eventually be able to build complete reliable AI systems. I see this as a huge transformation in our era. The mass production of digital robots can bring in huge transformation in our world.

Michael: Deeper modeling, with advances in unsupervised learning over unstructured data that incorporates all the ‘omics information datasets.

Olexandr: All current advancement in DL are primarily originating from breakthroughs in 3 field: image, speech and text recognition. I really excited to see emergence of DL in genomics, bioinformatics, systems biology, etc. I hope that these fields will pick up and drive the innovation. I keen to see what are new architectures and algorithms will appear next.

These experts will be speaking at the RE•WORK Deep Learning in Healthcare Summit in London on 7-8 April 2016. Other speakers include Brendan Frey, President & CEO, Deep Genomics; Daniel McDuff, Principal Research Scientist, Affectiva; Ekaterina Volkova-Volkmar, Researcher, Bupa; Alex Zhavoronkov, CEO, Insilico Medicine; Cosima Gretton, Doctor, Guy's and St Thomas' NHS Foundation Trust; Alejandro Jaimes, CTO & Chief Scientist, AiCure, and more.

Tickets for this event are now limited, for more information and to register please visit the event page here.

Big Data Machine Learning Deep Learning Algorithms Personalised Medicine Deep Learning Healthcare MedTech Genomics Deep Learning in Healthcare Summit Roundtable Q&A


    As Featured In


    Partners & Attendees

    Ibm watson health 3.001
    Rbc research.001
    Mit tech review.001
    Kd nuggets.001
    Maluuba 2017.001
    This website uses cookies to ensure you get the best experience. Learn more