Cognitive Computing In Medical Field With Applications

Abstract

This paper presents the importance of Cognitive Computing in real life and some important applications of Cognitive Computing in Medical Field.

Keywords—cognitive computing, data mining, natural language processing, I

Introduction

To Cognitive computing

Cognitive Computing is one of the most researched topic of 21st century. It finds its applications in many domains involving Finance, Artificial Intelligence, IOT and Medicine. Cognitive Computing uses various complex techniques to gain data from the environment where it is embodied. It not only interacts with the environment but also takes decisions just like a human beings. It’s a special application which uses self learning techniques like data mining, pattern recognition and Natural Language Processing. Using such complex techniques requires with a lot of data for a computing machine to take complex decision that a human brain is capable to take. A Cognitive agent is well designed to gather information from the surrounding environments. It collects processes and makes use of data as and when required.

Requirements Of An Cognitive Agent

Adaptive

The Cognitive Computing agent needs to a adaptive to the changes in the environment. It should be able to detect minor of all changes happening in the environment to make a better decision. They should be able to grasp information in any format like text, numerical data or even images. InteractiveThe Cognitive Machines have been designed to tackle a problem of bringing in the power of human thinking into machines. The main advantage humans have over machines is the ability to interact with the society. With developments, scientists are able to develop a complex brain-like machine like ‘Cognitive Machines’. Providing them with a way to interact with the society will make this machines more realistic. Hence Interaction is a necessity.

Stateful Cognitive Machines are so much powerful that they can sense a situation they are kept into and remember if ever again they face any similar situation. They have this powerful technique of analyzing the situation and take appropriate steps. Also the reaction to the situation is immediate and accurate. Just like the human brains, they have artificial neural networks which can fetch process and generate output for any situation. The outputs of these machines are very much similar to what human brain may have predicted if both face similar situation. Contextual Every decision ever taken by a cognitive machine need to dependent on a lot of factors. The answer depends on the place where machines are operating, the time when they are operating, the goals that user wants from the cognitive machine, the prior information this machines have, their computing speeds and the problem they are referring to. The machines are trained and designed in such a way that they can tackle any situation with just the same capabilities that a human brain can perform.

Different Applications of Cognitive Computing

Cognitive Computing has many applications in the real world. These applications are divided into different sectors of the work. Many of them include Finance, Medicine, Healthcare, Marketing, Robotics and Engineering. Some of the examples include IBM Watson, Vantage Software, Welltok, Lifelearn, WayBlazer, Edge Up Sports and Bright Minded. I would like to take one particular domain of the Cognitive Computing. It is the medicine and healthcare domain. Medicine is one of the vast domains of the 21st century. Many new great and innovative inventions are being created in this field. With development of new and better technologies, scientists are capable of finding new equipments for the betterment of society. Cognitive computing has widespread applications. One of them is in the Medicine and Healthcare Sector. The main need for a cognitive computing agent. Cognitive Machines In HealthcareImagine a situation. A child is sleeping in his room. Suddenly he faces problem in breathing due to Asthma. We have a cognitive machine install that can diagnose and take appropriate recovery steps. This machine can easily diagnose the problem by checking for pulse rate and heartbeat rate continuously at regular intervals. With sudden change in the rate, it can easily detect change in the rates and identify the problem.

Also it can take precautionary measures, like opening the window for fresh air to come inside. Also if precaution doesn’t work, it can signal to others about the problem. Next day, parents would be shocked to know that an Asthma attack was prevented. One of the main technologies being used to develop such a machine is to understand and interpret human behavior with the surrounding environment. Cognitive Computing is one such technology. It continuously monitors the human brains and makes complex decisions just like a brain does. A cognitive machine uses machine learning technologies, patter recognition, natural language processing and human senses processing based on real time acquisition of data. Cognitive Computing has an ability to perform extensively well in pattern recognition for individual patients using suitable sensors to understand patient’s conditions and also take correct precautionary measures whenever required. Some clinical trials with cognitive computing proved that the clinical benefits of the patients were improved using the real time operating cognitive clinical equipments as compared to a Clinical Assistant.

Further research is going on in this field. Oppourtunites for Cognitive Computing in MedicineWith advent of new technologies, new inventions are being made in the medical field. New ways are being developed to diagnose disease and monitor patients. There are few important domains where cognitive computing can help a lot. Medical Image AnalysisThe main technique used in image analysis is deep learning. It is a machine learning technique where the computer remembers the past activity and uses that to make correct decisions. For a new image to be analyzed, it will use the knowledge of the previous similar cases to give accurate analysis. Different devices are built which monitor different types of report like X-Rays, MRI and Medical Tomography. Some Machines focus on a specific organ like lung or intestine. They feed the machine with millions of images of same organ. This helps to identify an organ with some disease.

Medical Computations

A complex decision making machine requires a memory with sufficient RAM to perform complex decision making tasks. To implement deep learning and machine learning, we need to provide the Machine enough data using which it can predict the correct outcome. In order to diagnose any disease, it first requires enough information about the disease, its symptoms, how to cure it, finding out how much worse are the disease and medicines and medical equipments required to cure the disease.

Data Gathering

Data is one of the most crucial part of any cognitive agent. Data forms the basis of any cognitive machine. Different data gathering techniques are available to gain required information from the environment. Various new techniques and methods are available to gather and process data. As data is the primary and most important source of information. Cognitive machines should be such that they can gather rarest of data in any possible format. Also data should be stored inside the machine in such a way that it acquires minimum space and can be easily accessible. Also systems should be capable enough to acquire data in any form. For a successful machine, it must understand human language and interact with humans using that language. Data is gathered in an unstructured format by the device. It is this raw and unstructured which provide exact and accurate information.

Cognitinve Computing and Healthcare

Cognitive Computing helps doctors and clinicians make available a person’s healthcare data reports on a single go, which were earlier not available so easily and efficiently. Also each healthcare report for each and every person requires enough space to be stored. Gathering the healthcare data requires real time monitoring of a person’s activities. Data is to be read and acquired from within the body, for example heartbeat count, pulse count, number of breathing sessions per minute, number of calories gained and burnt, current body temperature and many more critical points. The main aim towards Cognitive Computing and Healthcare is to easy the process of real time information storage amongst all the persons. They would not have to be dependent on different devices for different parameters. They all will have a Cognitive Machine which will do all the tasks generate values for all the parameters required and update it regularly as and when required.

The data in the Healthcare Sector is widespread and exponentially growing. It includes data from various sources including the electronic healthcare report, clinical research, pathology reports, lab results, radiology images, voice recordings and other extraneous data. A cognitive machine interprets the data from all available sources in order to make an accurate decision. A cognitive machine with well equipped techniques and data can help the doctors and researchers to carry out proper diagnosis taking into consideration all the parameters. This vast data is sufficient enough to tell the all the medical facts of the patient. Once learned, it will remember the outcome and will use it in future predictions. Using the programmable computability, it can easily read and analyze data at a very fast rate. The algorithms used are themselves self –learning. With sufficient training, they become capable of modeling and analyzing any scenarios. The goal behind Cognitive machines in medicine and healthcare is to make medical diagnosis and patient checkup human free.

Advantages

Cognitive computing enables researchers to discover new insights in relationships among the genes, proteins, pathways and prototypes. Cognitive Machines help to identify most critical attributes of a patient and can provide easy and effective precautions and advice to patients. With training and ageing, machines become much powerful to detect any type of disease and cure them from root. Using the unstructured data from the Cognitive Machines, Radiologists can easily find the exact location if the problematic part. Cardiovascular Disease requires a lot of tests to be diagnosed. With Cognitive Computing, only one report from the Machine is enough to tell the nature of the problem with appropriate solution.

Case Study

With recent improvements in the IT industry in fields of Natural Language Processing and Cognitive Science, a so-called ‘e consultant’ has turned into a reality. One such device is the IBM Watson for Oncology. IBM Watson was launched on April, 2015. It provided an interface to collect necessary information in all possible formats and use it as and when it is required. Along with the device, they also created a cloud, named Watson Health Cloud, which allows information to be deidentified, shared and combined data with dynamic, constantly growing and aggregated view of clinical, researchIt is a ‘Memorial Sloan Kettering Cancer Center’ trained cognitive computing system which provides oncologists with ranked, evidence based treatment options for cancer treatment. There are 3 categories of treatments: ‘Recommended’, ‘For Consideration’, and ‘Not Recommended’. When the clinician gives a clinical question as an input the IBM Watson, the machine responds with a list of Hypothetical Predictions. This helps the clinician a lot in making a correct decision. IBM Watson has vital knowledge about the cancer diagnosis process and can use this knowledge with real time patient behavior to generate correct hypothesis about the cancer and appropriate steps to be taken for improve the patient’s health. This device was tested to test the outcomes that it generated while given the reports of around 300 cancer patients were fed to the machine to predict the analysis. The test reports show that for around 90% of patients, the results predicted by IBM Watson were very much similar to the actual measures taken.

According the test reports, there were still some places where the IBM Watson needs to improve itself. There were 10% cases where IBM Watson generated inappropriate output. This is of a very great concern for accuracy in medical field. If the initial data about the patient is not available in the IBM Pre defined format, no hypothesis is generated. Also, for few cases, it generated very different output. For example, the clinician recommended concurrent chemo radiotherapy for an 80 year old patient who could not endure radical surgery. But IBM Watson suggested radical surgery for the patient. The test concluded that IBM Watson can detect cancer and provide precautionary measures with an agility, accuracy and confidence. It could suggest a path after analyzing a complex situation. This can lead to faster treatment of the cancer patients. This was one of the cases where IBM Watson was put into use. There were several other diseases where other versions of Watson were used and the results of the test were satisfiable.