1) The AI Doctor

We are at a point in history where all the hype is about big data, machine learning, neural networks and artificial intelligence, and we are trying to integrate these concepts into basically every aspect of our lives. So how do we combine this with Healthcare? Let's have a look at where we are standing now.

Image source:https://tinyurl.com/y8a6nf6r

IBM Watson Health

IBM's Watson is a cognitive AI system developed with high emphasis on natural language processing. The system can understand unstructured data (which is 80% of the data produced by humans), and understands context .

IBM Watson Health is a commercially available cognitive system meant to assist doctors in decision making. It is delivered though the cloud, and was trained by analysing thousands of pages of literature to build of Corpus of knowledge built from data curated by humans.This corpus in 'ingested' resulting in a decision graphs, and then trained by experts with labelled training data. The process behind IBM's Watson is supposed to simulate the way humans think. Recently, the system has been extended not only to textual data, but also to imaging, to deal with eye, brain, breast, heart and related conditions[1]. The system returns evidence-based answers to questions, and it is continuously updating, by learning from its interaction with doctors.

Image source:https://tinyurl.com/ycddbtb6

It is considered that AI systems in healthcare will speed up the whole process, starting with data input by patients themselves, before meeting a health professional.[2]

source2:https://tinyurl.com/yb7pbvqg

Watson for Oncology

Watson for Oncology can help Oncology specialists to determine a possible solution for patients more quickly and efficiently. The system can analyse a patient's file, show the specialist the relevant information, and give options for treatment, whilst presenting also the reasoning it used to reach the specific solution. Moreover, this system can also be a help for patients to better understand their conditions and the treatments awaiting them.[3]

Another use of this system can be in board meetings, when different specialists meet to decide on the best treatment options for a patient. The system is already being used for this purpose. Dr. Wan Oijen expressed his opinion on the system:" it is the ideal additional discipline, providing up-to-date support during the tumor board meeting, but can also triage (prioritize) on beforehand the order in which patients are discussed during the tumor board meetings. External experts may then only be consulted for the first patients, which would result in a reduction of costs and efforts."[2]

Another help for the specialist is the fact that Watson is a constantly learning system, which keeps up to date with new data, guidelines and procedures, and it has easy access to metrics that can be used in the proposed suggestions.[3]

The system is being used in some hospitals in the US and in a hospital chain in India.

Watson Genomics

Since 75% of cancer patients do not respond to a particular kind of drug, a genomic-based approach has been introduced in cancer treatment. After a tissue sample of a tumour has been analysed and the genomic sequence was uploaded to Watson, personalised treatment options can be retrieved to determine the best possible solution, in a significantly lower amount of time, and by using a lower amount of resources. [4]

IBM Watson Paths

In order to simplify the access and understanding of EMD(Electronic Medical Data) for specialists, IBM developed Watson Paths, a system that displays medical unstructured and structured data in a more natural way. The system allows the study of case studies, medical records and reasons behind decisions.[5]

Google DeepMind Health

Recently, Google's DeepMind has received great publicity, when the AI Model has taught itself to walk.

Another use of their system has been in healthcare. Goole is working with Moorfields Eye Hospital NHS Foundation Trust to develop their system, and this month (July 2017) they have published an independent review and the changes that are to be implemented as a result (both can be accessed in [6])

Currently, the system is used in a hospital in London to alert specialists on patients' heath status. Nurses reported the system is saving them around 2 hours a day.

Careskore

Careskore is a cloud-based predictive analytics platform that aims to reduce mortality by running an analysis on patient data to determine their possibility of need of readmission. The system checks laboratory tests, behavioural data, demographics and it can inform doctors and patients of possible risks.[7]

Zephyr Health

This system uses global health data to aid the faster development of therapies. This system has a rather commercial orientation, helping Pharma companies find doctors with higher impact.[8]

Oncora Medical

This system provides predictive analytics for personalised radiation treatments of cancer patients. [9]

Deep Patient

This system uses structured text in the form of Electronic Health Records to extract patient descriptors that would help predict data for future patients.[10]

Image source:https://tinyurl.com/ya7dow8y

Sentrian

Sentrian is an FDA-approved analytics engine that harvests patient data from biosensors and then monitors the data from the sensor to alert patients and clinicians of possible upcoming problems. The aim of the system is to prevent the need of hospitalisation.[11]

For most of these systems, there is a common need: the need for learning data. This is the domain where IBM excels, as it has either collaborated with health institutions, or simply bought medical data companies. However, other ways of getting this data would be from patients themselves, by them filling out information from their clinical visits and providing information from health-monitoring devices. [12]