Requirements Model for Multi-Agent Autonomic Fetus Monitoring System

Ubiquitous Healthcare system provides the environment where users can receive medical treatment remotely at any time, any place. The purpose of u-healthcare monitoring system is to provide convenient healthcare service to patients and ease to diagnose patient’s health condition based on monitored physiological data for physicians. These devices perform many computing tasks like monitoring data, diagnosing vital signs remotely. The patient is equipped with one or more wearable devices which are attached to his/her body. The wearable devices have body sensors. The body sensors are connected with mobile apps through mobile gateways. In this paper, a fetus monitoring system is presented using wearable technology and U-healthcare. The wearable device is equipped with autonomic multi-agent software to monitor fetus’s and mother’s body parameters. Some of the parameters which can be observed are fetus heart rate, fetus body movements, mother’s BP and sugar level as well as contractions. The system generates alerts when body parameters of a mother and fetus change abnormally. If the parameters are changed drastically, the system sends the alert message to gynecologist of lady and also to family caretaker. This approach helps to keep watch on fetus development remotely since monitored data will be stored at the backend of the system. It reduces hospital trips and saves time and efforts of hospital staff for monitoring mother frequently [2][3]. The aim of this research paper is to represent requirements of Fetus Monitoring System in proper requirements models for generation of design phase models as well as development phase. This system uses PASSI technology (Process for Agents Societies Specification and Implementation) which is a step-by-step requirement-to-code methodology for designing and developing multi-agent system. The development of software using PASSI methodology is iterative and incremental [4]. This paper creates system requirements models which identifies domain for the system, identifies agents and task of agents. The research work generates requirements engineering models for FMS which consists of i) domain identification models to represent domain knowledge in suitable format, (use case diagram) ii) agent identification model to identify and represent agents of the system (package diagram) iii) role identification models based on scenario (class diagram) and iv) task specification model to represent logical capabilities of agents (activity diagram). AUML notations are used to generate these models. The requirements models are enhanced with case base reasoning with feedback loop to make the system adaptive. Case base reasoning cycle represents domain in the representation of cases. The MAPE-k loop identifies the scenario change, finds similar case from domain and plans for its implementation [6]. If no similar case is identified, it will be considered as new scenario or case and it will be added to case base as well as domain identification model is updated. These models help for further development of software.


I. Introduction
Due to diversity, heterogeneity and advances in technology, fetus monitoring, has made remarkable progress.Over the past few decades, electronic fetal monitoring systems have emerged as a very promising tool for use by midwives, obstetricians, and labor and delivery nursing staff.Further, wearable devices have provided unbiased, accurate data for monitoring patient's activity and this has enabled the healthcare community to monitor patient's outcomes and analyze overall trends.Chuan-Jun Su and Ta-Wei Chu [1] have presented a mobile system for fetal monitoring, however, it does not monitor the fetus from the first trimester.This paper presents a wearable device equipped with autonomic multi-agent software to monitor fetus's and mother's body parameters when mother is busy in her work, right from the first trimester of conception.This devise will help the working mothers to monitor fetus, It will monitor all the activity till the device in ON.The device will monitor fetus movements and heart beats.It will also generate alarms after regular time interval for water and food intake.Traditionally, doctors and nurses could only periodically and physically monitor the baby's heartbeats using

Requirements Model for Multi-Agent Autonomic Fetus Monitoring System 4th -Somaiya International Conference on Technology and Information Management (SICTIM'18)
2 | Page K J Somaiya Institute of Management Studies and Research (SIMSR) sonography or colour Doppler methods.This approach did not allow them to detect changes in fetal heartbeats at any time or offer continuous surveillance.Thus, physicians still lacked the important information for reducing perinatal or neonatal mortality rates.The proposed device will help physician to monitor fetal safety.When the water level goes down in mother's body, or the baby does not show any movement for long interval of time, the device will generate alert message for mother.So she will be alert and may drink water to maintain appropriate water level.If mother's water level or fluid level become very low or BP level becomes abnormal then monitoring system will detect and record the changes and send alert message to home caretaker and obstetrician.So they can reach patient immediately or suggest medicine on call.If the emergency index is very high then the monitoring system can also track the nearest ambulance to send patient to hospital.
The monitoring data will be stored as patterns.The collected data patterns will be useful to determine the status of fetus during prenatal stage or birth process.Such fetal remote monitoring system will allow pregnant woman to track fetus development as well as it will reduce the number of hospital trips.If some abnormal patterns of data are detected, then the obstetrician can suggest medicines to pregnant woman on call.If the abnormality is severe, then the obstetrician may tell her to visit the hospital.The proposed system is useful in the scenario when the pregnant woman is engrossed in her work and forgets to drink water or take food or forgets to keep a watch on baby movements, especially in second and third trimester.The agent at front end is wearable technology which will monitor and generate patterns.These patterns will be stored in storage.At the backend there will be mobile app which will be connected to patient's remote mobile application [2].
This paper performs analysis of multi-agent autonomic fetus monitoring system and identifies agents and interactions among the agents for the system.The major contribution of this study will be to proposecase base reasoning for making the monitoring autonomic by consideringreal-timescenarios.Thus fetus andpregnant woman's safety is improved and medical staff is relieved from tedious data monitoring task and during repeated hospital visits.
Section 2 will define PASSI (Process for Agents Societies Specification and Implementation) methodology and case base reasoning.Section 3 describes requirements models for fetus monitoring system.Section 4 explains design implementations of requirements models.Section 5 concludes the research work.

PASSI:
Massimo Cossentino [3] Proposes Process for Agents Societies Specification and Implementation.It is a step-by-step requirement-to-code methodology for designing and developing multi-agent system.The development of software using PASSI methodology is iterative and incremental as shown in figure 1.
During systems requirements modeling, the agent is autonomous entity which can take decisions, implement actions and has social relationships with other agents in organization.The agent performs sequence of tasks as a role.When agent implements some behavior in collaboration with itself or with other agents, it will be considered as a task.Each agent's behavior is mentioned in ontology.
Here PASSI (Process for Agents Societies Specification and Implementation) methodology is applied for requirements engineering of mentioned system.According to PASSI process flow, during requirements engineering activities executed are shown in figure 2 [3].

Requirements Model for Multi-Agent Autonomic Fetus Monitoring System 4th -Somaiya International Conference on Technology and Information Management (SICTIM'18)
3 | Page K J Somaiya Institute of Management Studies and Research (SIMSR) Source: Model-Driven Engineering of Adaptation Engines for Self-AdaptiveSoftware: Executable

Case Base Reasoning with MAPE-k loop:
Case-BasedReasoning resembles human reasoning where symptoms represent problem, diagnosis and treatment represent asolution.CBR collects cases where each case consists of a problem, solution and its result.The CBR cycle is as shown in figure 3. The CBR cycle consists of 4 steps which are performed on case base.i) The Retrieve phase selects one or several similar cases from the case base.ii) In Reuse phase, historical similar cases are decided to be implemented.iii) In the revised phase, the matched solution is verified according to environment and corrected or improved if required.iv) Finally, the retain phase takes the feedback from the revise phase and updates the case base.Thus, CBR either converts new cases in learned cases which will be available for future decisions or finds similar historical case and reuses it [4].[6] designed the architecture for self-adaptive system with autonomic element.The autonomic element consists of managed element and autonomic manager.The managed element is adaptable software and autonomic manager is adaptation engine to execute MAPE-k engine.MAPE-k engine consists of monitor to sense the managed process, its context and stores relevant events in knowledge base for further reference.The analyzer compares scenarios with patterns in knowledge base to diagnose symptoms, the new symptoms can be stored in knowledge base.The planner interprets the symptoms and designs a planto execute change in managed process.The executor executes plans to adapt actual system and obtain the desired output.The sensors and effectors collaborate data and control among autonomic elements.MAPE-k cycle is called feedback-loop used for engineering self-adaptive software system.The feedback loop ensures that the adaptation engine helps the adaptable software to respond change in requirements from the environment [5].Feedback loop helps self-adaptive software system approach from analysis design till implementation.
Case base reasoning can be incorporated with MAPE-k cycle for adaptivity.As shown in fig 3, case base reasoning consists of case base which represents all cases learned by the system.The case base can represent domain knowledge or ontology for the system.Using case base, the analyzer of MAPE-k loop can analyze the scenarios and if new scenario occurs then it will be compared with existing scenarios.The scenario is converted into case and case base is updated if completely new scenario is identified i.e. retain process is followed.The planner plans the new scenario with the reference of new case.The implementation of MAPE-k cycle will be easy with case base reasoning.

Existing Systems for fetal monitoring:
Hammar and Hewlett Packard developed first commercial electronic fetal monitoring device in 1978.This device was used to monitor heart cycle, heart sound during labor [2].Jansuz developed fetal heart rate signal processing.Czabanski, R.; Jezewski, M.; Wrobel, J.; Jezewski, J.; Horoba, K. developed artificial neural network to evaluate risk of low fetal birth weight using cardiography signals.New innovations are going on for electronic fetal monitoring using wireless communication technologies, portable monitoring devices like RFID.Fetal data can be transmitted to remote locations via wireless communication devices [2][7].
A research group of Sandy Pentlandidentified location, movement and communication data from smart phone to identify flu and gastrointestinal problems.William Kaiser from UCLA developed healthcare software for stroke patient monitoring and guidance for rehabilitation over 20 countries [8]. Location Tracking Services: With the help of GPS, location estimation mechanism, location tracking services can be developed for elderly, visually impaired and intellectually disabled patients [7]. Sleep Analysis: The patient sleeping patterns can be monitored using gyroscope and magnetometer.These devices are embedded into smart phones.The sleeping behavior is monitored by strapping the mobile phone on arm or on waist or keeping mobile below the pillow [7]. Requirements

III. System Requirements Models For MAA-FMS
When the pregnant woman is busy in her work, she may neglect her water and food intake.In critical pregnancy cases, heart rate, BP or uterine contractions or baby movements should be watched continuously.The illiterate pregnant woman may not know the quantity of water, food she has to take to grow fetus properly.If water intake is not in adequate quantity, the fetus may dehydrate.Inadequate food intake can cause malnutrition of fetus.It is practically impossible for obstetrician to keep watch on daily intake of water and food of mother as well as baby movements for 24* 7. Hence, it is essential to develop a system which works with wearable device and generates alerts device will generate alert for caretaker and obstetrician of the patient.
To satisfy these requirements, this research proposes Multi-Agent Autonomic -Fetus Monitoring System (MAA-FMS)in which wearable device will collect monitoring data and spontaneously inform pregnant woman for intake and caretakers about vital signs.The MAA-FMS consists of two parts: first component is front end MAA-FMS which will be the wearable device which pregnant woman will wear.Second component is backend MAA-FMS which will be with caretaker agent and gynecologist agent.The functional view of the system is as shown in figure 5.
The fig describes that MAA-FMS monitors the fetus and mother body paralmeters.When the parameters are chnaged drastically, the backend of the system will check for any emergency.This is supported by case base and MAPR-k loop.If the scenario is same as previously recorded scenario then the predecided actions are taken and alert is sent to gynecologist.If the new scenario, then immediately alert is sent to gynecologist.For this system based on functional requirements identified, following requirements engineering models are generated.

Domain Requirements Description Phase:
This phase describes functional description of the system in hierarchical series of use case diagrams.Two perspectives can be considered to describe domain requirements for MAA-FMS.One perspective is according to mother where she will make the wearable device "ON" before sitting on work desk.The device will start monitoring food and water intake for gravida and fetus.If the diagnostic agent senses some abnormal parameters, it will inform wearable device and generate alerts for mother and caretakers.All monitored data will be stored at knowledge -based data server so that whenever obstetrician needs data, it can be downloaded.The domain description model for mother's perspective is as shown in figure 6.

Fig 4
describes MAPE-k (Monitor, Analyze, Plan and Execute-Knowledge) Cycle.Kephart and Chess Model for Multi-Agent Autonomic Fetus Monitoring System 4th -Somaiya International Conference on Technology and Information Management (SICTIM'18) 4 | Page K J Somaiya Institute of Management Studies and Research (SIMSR)  Vital Sign Monitoring: There are iPhone add-on modules and mobile apps to record heart rate, ECG, blood pressure and blood sugar using Personal Area Network.This data is stored as patterns and can be useful for physician when needed [1][2].