US-China Education Review A 6 (2012) 619-626 Earlier title: US-China Education Review, ISSN 1548-6613

D

DAVID

PUBLISHING

Mobile Phone Based Medical Diagnostic System Derrick Ntalasha The Copperbelt University, Kitwe, Zambia

This paper discusses a mobile phone accessed expert system that is used for diagnosing diseases. It is called the cell phone based medical diagnostic system. This system provides ease diagnosis of diseases even in areas where there are insufficient medical staff. The system provides automated help through a cell phone for rural hospitals and those hospitals which are under staffed. The system is used to provide automated help by storing knowledge of diseases, this includes diagnosing a particular disease and the necessary measures to be undertaken, when queried about a particular diagnosis or symptoms obtained from the patient. The medium of access for the system is a cell phone (mobile phone) where a text message written is sent to the application. The system produces the desired output and sends the message back to the cell phone that sent the message before. The Java Expert Shell System forms the “brain” of the system that does all the reasoning of the system. Keywords: expert system, automated disease diagnostic, cell phone, text messaging, knowledge base, Java Expert Shell System

Introduction Our health care systems in Zambia and many developing countries have few experienced medical staff and no computerized systems in place to diagnose common diseases. This makes it hard for common people to access proper medical attention and care. The few experienced and de-motivated medical staff are overwhelmed with so many patients to attend to every time. These constraints, added together with poverty situations in our countries, have made it even more difficult for common person to seek medical attention. The solution to such problems is to developing a computerized system that automates the diagnosis process of diseases, especially in areas with poor medical facilities, lack of experienced medical personnel and where people do not have sufficient money to access proper medical attention and care. The mobile phone based medical diagnostic system is an expert system (Retrieved from http://www. en.wikipedia.org/wiki/Expert_system) for diagnosing diseases using a computer which is located remotely. This system provides ease diagnosis of diseases as long as a user has a mobile phone which can be used to send a text message. The system provides automated help through a mobile phone. This system has an expert knowledge base (Adrian, 1993), which is used to diagnose a particular disease or symptoms sent by a patient. A patient uses a mobile phone to send a text message to the system which is used to search through the expert knowledge base. The search results into a system state the problem of the patient and suggest the medicines required for the diagnosed disease. This diagnosis is then sent back as a reply to the mobile phone that sent the text message. The cell phone based medical diagnostic system achieves the following objectives and goals: Derrick Ntalasha, Ph.D. candidate, M.Sc., B.Sc., Computer Science Department, The Copperbelt University.

620

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

(1) To provide expertise medical diagnosis services (that is, those that a professionally trained personnel will do) in areas that have under qualified medical personnel; (2) To empower everyone to an equal share of medical attention, so that even people in areas with very poor medical systems and lack of trained personnel, will be catered for as a cell phone can work effectively even in very remote places; (3) To provide a low cost of treatment service to people who have no money to pay for consulting professionally trained medical personnel. All that is needed is a cell phone to access such services from the system; (4) To provide instant medical attention as there is no need for appointments as is the case with the current manual system where you are required to make an appointment with a medical practitioner.

Related Works An expert system for health care administration is engineered to automate knowledge-based data used in the field of health care. A robust expert system provides input scenarios, problem-solving mechanisms and output data based on a review of common practices and hospital support functions. Expert systems can also emulate human behaviors and logic patterns to answer questions and solve problems. NxOpinion Mobile Data Platform Software, named NxOpinion Mobile Data Platform (Dedecker, Van Cutsem, Mostinckx, D’Hondt, & De Meuter, 2006), exists with a mobile platform to address limited portability of health information in resource-poor areas. It utilizes a cell platform with an integrated data management system and a Bayesian inference engine (Allen & Helferich, 1990). It has a diagnostic engine, a medical database and an integrated health record (Mohanand & Sultan, 2010). The database has more than 300 diseases with treatment information. Users can send data for monitoring, evaluating and following-up. Minimally trained healthcare workers can evaluate, diagnose and treat patients. It provides clinical supervision of healthcare providers in remote areas. The cell phone platform can initiate a patient health record stored on a central server for access by medical facilities. Immunization and national program monitoring can be accommodated. Alerts can be sent for follow ups or for immunizations. It provides statistical data and culturally specific regionalization/language support. Patients can control their healthcare records. It allows resource-limited systems to provide excellent medical care, better utilize available resources, and interact with regional and national public health initiatives (Permanand, S. Sultan, & N. Sultan, 2009). NxOpinion has been developed into a proven medical diagnostic decision support and a tool for gathering data in the hands of minimally trained health extension workers. The platform has been adapted to function on low end, readily available cell phones for use in remote and low resource settings, providing critical and connected medical infrastructure. MediNet MediNet is a mobile healthcare system for patients in the Caribbean suffering from diabetes and cardiovascular disease (Qureshi & Tounsi, 2006). The first phase of the project was funded by a grant from Microsoft research (“cell phone as a platform for healthcare RFP (Request for Proposal)”), and resulted in the development of a local MediNet for Trinidad and Tobago (Mohanand & Sultan, 2010). This system is currently being tested, and preparations are under the way to conduct a field test of the system in several

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

621

Caribbean countries in the near future. The main objective of the system was to network the healthcare resources in the Caribbean region with the aim of providing personalized healthcare services to a wider population regardless of economic status or location. The experiences in the training of patients to use MediNet and present statistics, showing usage of the system during the testing period, will be conducted. The MIRHMS (Mobile Intelligent Remote-Access Healthcare Management System) The MIRHMS are typically designed to integrate a medical knowledge base, patient data and an inference engine to generate case specific advice (Cheung, 2008). The MIRHMS solver incorporates knowledge bases that represent the medical and clinical model used by medical doctors. The solver uses some medical and symptoms parameters collected by the wireless devices and/or sensors. Thus, the solver evaluates the patient’s medical status using the collected input data.

The Cell Phone Based Medical Diagnostic System CBMDS (Cell phone Based Medical Diagnostic System) allows easy diagnosis of diseases even in areas where there is insufficient medical staff, prominently in rural areas. The system is aimed at providing online help through a cell phone for most of the Zambian population, but it concentrates more on rural people or those hospitals which are under staffed. The system provides online help by storing knowledge of diseases and diagnosing a particular disease and the necessary measures to be undertaken. When the system is queried about a particular diagnosis or symptoms obtained from the patient, it provides the user with the procedure and advice on how the disease should be treated. The following are the functions that the CBMDS will deliver: (1) The system allows a user (patient) to submit his/her symptoms of the disease in form of SMS (short message service) which will be interpreted by the inference engine; (2) The system is able to store messages in case mobile phone line is busy or out of order. These messages are a combination of symptoms for a certain disease that that the user of the system sends through the cell phone; (3) The system generates an appropriate diagnosis depending on the combination of symptoms input; (4) The system gives the prescription for each diagnosis performed. The following sections give a detailed description of the data components of the system. Each section specifies the features that have been incorporated in the system and how they were developed, such as file descriptions, data types, passwords, the user interface and modules for each activity in the application and a description of program modules. Data Design Data design specifies all the input, output and stored files that the system expects. This gives a specification on the structure of the data that should be input into the system and the output that is expected of the system. This gives the data type, length and a description of each data item. Input data. This is data that are input into the system. This information includes the symptoms the patient submits for a particular disease in order to be diagnosed. This information, symptoms of the disease, is input in form of SMS with the GSM (Global System for Mobile Communication) modem, acting as an interface between the cell phone and the computer, where the expert shell is housed. Administrative data. This is the data that an administrator is expected to supply to the system. The administrator is one who is responsible for the inputting of symptoms that the system is expected to provide

622

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

diagnosis for. An administrator can also edit the knowledge-base, if there is a need. Disease data. The disease data are stored as disease name and the various symptoms attributed with the same disease. This information is embodied in the expert shell’s knowledge base as facts and rules. A rule is basically represented as an ordered pair (, ) or as: If , then . The or the “if” part describes a certain situation represented by a set of facts. The or the “then” part describes a new situation represented by another set of facts (Jean-Louis, 1999). Figure 1 shows the structure of the rule (facts) based system. Inputs

Rules and facts

Working memory

Selected rule

Outputs

Rule interpreter Selected date

Rule and data selection

Figure 1. Rule based system architecture.

Output Data. The output, which is a result of a combination of symptoms of a particular disease, is the information about whether or not a patient has a disease that the system provides diagnosis for. In addition, the system provides the necessary prescription for a given combination of symptoms. The output, like the input, is displayed on a cell phone in form of a text message. The cell phone in this case acts both as a sender and receiver Application Design The logical design provides the flow of logic in the system. It describes the design of an application giving the modules and shows how the modules interact to accomplish the system’s objective/goal. Figure 2 gives the flowchart showing the flow of logic in the CBMDS and the logical design of CBMDS. Input SMS module. This module facilitates the input of symptoms of a disease through SMS. SMS in correct format module. This validates whether the correct format of an SMS has been adhered to during the input process. If the format is incorrect, the “INVALID SMS FORMAT” notification is displayed else the check SMS in GSM modem module invoked. The format of the SMS is in bulk form (bulk SMS) where a group of symptoms delimited by commas is used, e.g., headache N\Y, temp 32, pain_in_joints Y\N, fever Y\N, nausea Y\N and send it to the number of the SIM (subscriber identification module) card in the GSM modem. The sequence in which the symptoms are to be entered is such that the order conforms to the order in which rules are defined in the knowledge base. For example, if the rules are defined such that headache is first then followed by temperature, pain_in_joints, fever, nausea, etc., then the symptoms should also be input

623

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM in the same sequence. Start

Input SMS

SMS in correct format?

“Invalid SMS format”

No

Yes

Rule-base

Check if SMS in GSM modem

Notify User “SMS NOT DELIEVERED”

Not

Present To Expert Shell Retrieve SMS

Loads SMS (facts) into the working memory

Facts and rules

To mobile phone

Display SMS on cell phone

Diagnosis by inference engine Execution engine

End

SMS sent back to modem Figure 2. CBMDS logical design.

Check SMS in GSM modem module. This module checks whether the SMS has been successfully sent to the GSM modem. If an SMS is not in the GSM modem, the system displays a notification message “SMS NOT DELIEVERED”. If present, it invokes the retrieve SMS module. Retrieve SMS module. This module retrieves an SMS from the GSM modem, which is used to query the inference engine or for display on the cell phone. Working memory. Facts in form of SMS that are loaded in the working memory are forwarded to the inference engine for interpretation. Query the inference engine. This module interprets the facts that are entered in from SMS. To achieve this, the inference engine uses stored system rules from the rule base and the execution engine. SMS fired back to GSM modem. After querying with the inference engine, the system sends an SMS back to the GSM modem (Retrieved from http://www.nowsms.com/faq/what-is-a-gsm-modem/), which is a result of the SMS, querying with the inference engine.

624

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

Display SMS on cell phone. This module displays the result of query with the inference engine by sending it to the cell phone. Physical Design The physical design shows the physical components that constitute the system. It provides the physical layout of the components. When the cell phone based medical diagnostic system application is run, several processes take place as illustrated in Figure 3. The discussion outlines how the physical design supports the system’s logical design to meet the user’s requirements. Expert Shell Knowledge base Working memory

Rules and facts

GSM modem Selected data

Rule interpreter

Send SMS using a mobile phone

Rule and data selection

Selected rule

Figure 3. Physical design of CBMDS.

Mobile phone. The text message is composed and sent to the computer which houses the inference engine. The computer (inference engine) then sends the message back to the mobile phone, and this is read as a text message. The medium to act as an interface between the mobile phone and the computer is the GSM modem. To achieve this, the message is sent to the number of the SIM card that is in use in the GSM modem. The mobile phone serves as an input and output device, sending and receiving text messages. It is only the sender’s point of contact with the system. GSM modem. The modem is connected to either the computer’s communication or universal serial ports. The received message is temporarily stored in the modem before being extracted by the system. The connection between the mobile phone and the GSM modem is wireless, and the connection between the computer and the GSM modem is physical, which is through the USB (universal serial bus) port. Working memory. This is also called the Work Space. This plays the role of “short term memory” for all messages that need to be exchanged between the application the cell phones in use. The knowledge base. This stores the permanent knowledge of the domain of the application and allows the system to act as an expert in the domain under consideration. It is this module which depends on the domain of application. It holds a set of rules of inference (production rules) that are used in reasoning. These rules form the form; If , then . The knowledge base contains knowledge belonging to the domain of the system; it stimulates the activity of an expert in his/her deductive and explanatory capacity. The rule interpreter or inference engine. This controls the overall execution of the rules and provides the reasoning ability that enables the expert systems to form conclusions. The inference engine works by

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

625

searching through the knowledge base and working memory, attempting to match facts present in the memory to the antecedent of rules. There are two types of inference Mechanisms, these are forward chaining and backward chaining. Backward chaining. Backward chaining is also said to be goal-driven, because it starts with the goal, which is the solution that it tries to verify (Russell & Norvig, 1995). In backward chaining, the expert system starts with the possible solution and attempts to gather information to verify the solution. If at any point it requires some information, which is not present in the knowledge base, it questions the user. The user’s answers to the question then become part of the problem specification in the working memory. Forward chaining. In this mechanism, which is also said to be data-driven, due to the fact that it starts with data (problem specification), the user starts off by entering all the relevant data related to the problem (Russell & Norvig, 1995). The system stores these data in its working memory. After this, the system does not question the user further. The inference engine uses the available data and chains forward to reach a conclusion. The application is implemented in a public class called CBMDS.java. It utilizes a C# SMS engine to send and receive SMS messages from the GSM modem. The CBMDS.java as well as defining the user interface for the system provides a link to the SMS engine. When a user sends a message, the incoming message to the system is stored in the incoming table of the jessdb database. The system implemented in Java then retrieves the SMS and then splits the SMS to eliminate the comma “,” that separates the various symptoms input. The split message is then loaded into the assert statement of the CBMDS.java. The system, then, using the ASSERT statement, determines which rule to execute using the pattern matching algorithm used in JESS (Java Expert Shell System) called the Rete Algorithm (Strauss, 2002). When the rule is fired, the appropriate response is stored in the outgoing table of the jessdb database from the SMS engine using the GSM modem which sends the SMS back to the cell phone that sent the message. Message Delivery A message is delivered on the sender’s cell phone. Therefore, there is no need to specify the recipient’s number, and the only number to specify is the number of the SIM card in the GSM modem. Usability Usability of the system refers to the ability of the user to interact with the system without any difficulties. The system provides great flexibility. It enables the system user from any location to get diagnosis. The user types in the text message in a valid format and then sends it to the number of the SIM card in the GSM modem. The system is also user friendly, as it incorporates a graphical user interface to ease its administration.

Conclusions The CBMDS provides automated help through a cell phone for individuals, rural hospitals and hospitals which are under staffed. The system provides automated help by storing a knowledge-base of diseases. The system diagnoses a person for a particular disease. This involves a patient, submitting the symptoms. According to the knowledge stored in the knowledge base, the patient receives appropriate diagnosis. The cell phone acts as a means through which the patient’s symptoms of the disease are input into the system, and the appropriate responses display accordingly. The cell phone provides the GUI (graphical user interface) for the system which users can use to perform the various tasks that are expected in the system.

626

MOBILE PHONE BASED MEDICAL DIAGNOSTIC SYSTEM

References Adrian, A. H. (1993). Knowledge-based systems for engineers and scientists (p. 17). New York: CRC Press. Allen, M. K., & Helferich, O. K. (1990). Putting expert systems to work in logistics. Intellogistics, Inc. and Dialog Systems Division of A.T.Kearney, Inc.. Cheung, S. L. (2008). Using mobile phone messaging as a response medium in classroom experiments. Journal of Economics Education, 39(1), 51-67. Couch, & Steinberg, D. H. (2002). Java 2 enterprise edition bible. 909 Third Avenue New York: . Hungry Minds, Inc.. Dedecker, Van Cutsem, T., Mostinckx, S., D’Hondt, T., & De Meuter, W. (2006). Ambient-oriented programming in Ambienttalk. In Proceedings of the 20th European Conference on Object-Oriented Programming (ECOOP). Nantes, France. Jean-Louis, E. (1999). Expert systems-theory and practice (p. 15). Prentice-Hall, New Delhi, India. Mohanand, P., & Sultan, S. (2010). Staying connected in a mobile healthcare system: Experiences from the MediNet Project. E-minds: International Journal on Human-Computer Interaction, II(6). Permanand, M., Sultan, S, & Sultan, N. (2009). Developing a mobile health system for patients in the Caribbean with diabetes and cardiovascular disease: Experiences from the MediNet project. International Journal for education and Development using Information and communication Technologies, 5(4). Qureshi, B., & Tounsi, M. (2006). A wireless sensor based intelligent mobile system for remote healthcare monitoring in Saudi Arabia. Proceedings of 1st NITS: Bridging the digital divide, challenges and solutions (pp193-198). Riyad Pfleeger, S. L., & Atlee, J. M. (2002). Software engineering, theory and practice (3rd ed.). Pearson International Edition. Russell, J. S., & Norvig, P. (1995). Artificial intelligence, a modern approach. Eaglewood Cliffs, New Jersey: Prentice Hall-Inc.. Sabat, M., & Alpert, J. (2009). Using SMS to aid flu vaccinations. The Journal of the American Medical Association, 3(5). Schildt, H. (2003). JAVA 2, A beginners guide (2nd ed.). McGraw-Hill. Steeb, W. H., & Dembskey, E. J. (2009). Programming in C#. International School for Scientific Computing. Strauss, M. (2007). Jess, the Java Expert System Shell: Artificial intelligence tools. Univerisity of Saarlandes. Valdez, J. M. G. (2002). Object oriented design and implementation of an inference engine for fuzzy systems (2nd ed.). McGraw-Hill. http://www.nowsms.com/faq/what-is-a-gsm-modem/ http://en.wikipedia.org/wiki/Expert_system http://www.iaeng.org/publication/WCE2009 http://www.en.wikipedia.org http://en.wkipedia.org/wiki/Mhealth http://www.howstuffswork.com http://papers.ssrn.com/sol3 http://www.codeprojec.com/csharp http://www.sciencedaily.com/releases/2009 http://www.springerlink.com/content http://www.wikipedia.org http://www.w3.org/2008/02/ms4d_ws/cdacmobile-healthecare-paper.pdf