Artificial Intelligence in Medical Science

"Artificial Intelligence in Medical Science"

From the very beginning of the era of computers, scientists were trying to create artificial intelligent systems such that robot will have characteristics similar to a human body. Among all the modern technological quests, people have been trying to build artificially intelligent (AI) computer systems for different sectors, which is one of the most challenging tasks in this field. And in fact, artificial intelligence has developed in different sectors, among which medical science is notable. In medical science the contribution and usefulness of artificial intelligence has developed tremendously and it continues to develop. Artificial intelligence in medical science has numerous applications, for instance, it is used to solve dermatological problems of people in rural areas, to identify emergency patients based on their symptoms, to protect patient data in different health organizations, to identify significant factors for diabetic patients to control their diseases and many more.

In a large country, it is hard to offer medical service properly to each citizen. Moreover, in rural areas it is very hard to provide an appropriate level of health care system. The number of medical specialists in rural areas is less than in the cities. Researchers have been trying to propose a decision support system to meet the needs of medical practitioners in rural and remote areas. A decision support system can be defined as the system in which the system will learn from the trained data, which stands for valid information and gathered from trusted sources, and then on the basis of that data the system will find out the results and will provide a decision about what should be done. A decision support system can serve as a solution for the treatment to people. A decision support system can be used over the Internet and even in rural areas by communicating through the Internet by using different methodologies, like video conferencing, telemedicine, and access to databases etc. Moreover, it is challenging to train general practitioners in rural areas with the decision support system in the field of dermatology. Ou et al. (2008) have built a decision support system, TELEDERM, to offer increased functionality to the general practitioners and to help them diagnose patients. In TELEDERM, a reasonable case base, which can be thought of as a database, is built up with dermatological cases. The case base database is stored with different types of questions and diagnoses that took place during diagnosis by general practitioners (GPs) and consultants. Finally, the TELEDERM system is further validated by a consultant. The TELEDERM has proved its efficiency in providing medical treatment to patients and it is accepted by most of the general practitioners. “Feedback from GPs concerning the usefulness of TELEDERM was positive with 67% finding the system useful or very useful, and 25% finding the system somewhat useful.” (Ou, West, Lazarescu, Clay, 2008)

A decision support system is also used in identifying the symptoms of diseases. By applying the decision support systems, it becomes easier for the doctors to find out the reasons behind a particular disease and also the source of it. Also, doctors can identify the emergency patient and can give treatment accordingly. “‘Machine learning’ or computer-assisted predictive models have been successfully utilized to optimize treatment and predict clinical outcomes in a variety of other conditions, such as computerized interpretation of the electrocardiogram, to help streamline and optimize care of patients with acute myocardial infarction, especially in a busy practice or in the emergency room.” (Chu 2008). For example, Chu et. al. (2008) have applied mathematical models to construct decision support systems for identifying the bleeding source amongst patients with acute gastrointestinal bleeding (GIB) requiring urgent healing and endoscopy. The mathematical model proposed by Chu et. al. (2008) has succeeded in providing suggestions for patients with success rate above 70-80%.

Preserving the patients’ data is one vital task in medical health organizations. When disclosing the data within a health care organization, the anonymity of the patient should be preserved. However, it is possible to identify patients’ data easily from the public information e.g. voter registration database. Also, “a patient’s location visit pattern, or “trail”, can re-identify seemingly anonymous DNA to patient identity” (Malin, 2007). To prevent trail re-identification, Malin (2007) has proposed a privacy protection model, k-unlinkability, which helps medical health care organizations to reveal DNA records for a particular patient that are unattainable to trail re-identification. Malin (2007) also said that “We developed several methods by which health care organizations can collaborate to share patient-specific biomedical data with provable protection from trail re-identification; i.e., the linkage of de-identified biomedical data to named patients via hospital-visit patterns.” The privacy protection model deduced by Malin (2007) is a technical solution to health care organization to eliminate trail re-identification. Moreover, it is required to make agreements between the health care organizations to exchange information. Malin (2007) said that “To ensure the adoption of our technologies and prevent trail re-identification, it is necessary to account for policies and draft appropriate agreements.” Actually, this research has solved the security problem with patients’ data in health care organizations with the help of artificial intelligence.

The number of diabetic patients in the world is increasing every year. According to the World Health Organization, it affects around 194 million people worldwide, and that number is expected to increase to at least 300 million by 2025. Diabetes may cause serious damage to health; this can even cause death if not controlled. Diabetic patients are of two types. The type-1 patients don’t have to take insulin. But, the type-2 patients have to take insulin and medicine to control the blood sugar in their body. There are several risk factors that can hamper diabetic patients. For type-2 patients, Huang et al. (2007) have applied the feature selection and classification model construction (FSSMC) to find out the major factors that influence diabetes control and to identify the population with poor diabetes control status. Patients’ ‘age’, ‘diagnosis duration’, necessity of ‘insulin treatment’, evaluation of ‘random blood glucose control’ and ‘diet treatment’ are the influential factors for blood glucose control with higher priority as discovered by the FSSMC model (Huang, McCullagh, Black, Harper, 2007). To find out the top 15 ranked features, FSSMC model is applied to predict patients’ diabetes control. (Huang, McCullagh, Black, Harper, 2007, 256) (Table-3). This table validated that age and diagnosis duration are the major factors over blood glucose control and the features listed in the table helps to select those efficiently and to extract valuable information from large medical databases. This study has solved a problelm that people were facing before. Huang (2007) have rightly


Huang, McCullagh, Black, Harper, 2007, 256

said that “the diabetics specialist was encouraged by the fact that the data-driven analysis in this study was able to verify by importance of rank much of the knowledge which has been researched for over 50 years in the medical domain using epidemiological studies, laboratory investigation and inductive reasoning by clinicians.”

Application of Artificial intelligence in medical science is growing every year. Researchers are working hard to make improvements in this field. Though they are facing complexities while working with this, acceptable results are bringing success in this field. It is being used to do diagnosis based on symptoms, to treat faraway patients, to preserve patient’s identity and many more. It is making life easier by helping to solve problems and also to take quick decisions. Poor countries will be benefited if artificial intelligence in medical science is developed properly. This will reduce the number of patients in a country in the future.


Chu, Adrienne, Ahn, Hongshik, Halwan, Bhawna, Kalmin, Bruce, Artifon, Everson L.A,

Barkun, Alan, Lagoudakis, Michail G., & Kumar, Atul (2008). A decision support system to facilitate management of patients with acute gastrointestinal bleeding. Artificial Intelligence in Medicine, 42, 247-259.

Huang, Yue, McCullagh, Paul, Black, Norman, & Harper, Roy (2007). Feature selection and

classification model construction on type 2 diabetic patients’ data. Artificial Intelligence in Medicine, 41, 251-262.

Malin, Bradley (2007). A computational model to protect patient data from location-based

re-identification. Artificial Intelligence in Medicine, 40, 223-239.

Ou, Monica H., West, Geoff A.W., Lazarescu, Mihai, & Clay, Chris D. (2008). Evaluation of

TELEDERM for dermatological services in rural and remote areas. Artificial Intelligence in Medicine, 44, 27-40.

© 2008 by Quazi Mainul Hasan. All rights reserved.


Popular posts from this blog

html to Word Document Converter using Open XML SDK

How to: Get Top n Rows of DataView in C#