Journal of Medical Systems

, 40:222

Health Monitoring and Management for Manufacturing Workers in Adverse Working Conditions

  • Xiaoya Xu
  • Miao Zhong
  • Jiafu Wan
  • Minglun Yi
  • Tiancheng Gao
Patient Facing Systems
Part of the following topical collections:
  1. Smart and Interactive Healthcare Systems

Abstract

In adverse working conditions, environmental parameters such as metallic dust, noise, and environmental temperature, directly affect the health condition of manufacturing workers. It is therefore important to implement health monitoring and management based on important physiological parameters (e.g., heart rate, blood pressure, and body temperature). In recent years, new technologies, such as body area networks, cloud computing, and smart clothing, have allowed the improvement of the quality of services. In this article, we first give five-layer architecture for health monitoring and management of manufacturing workers. Then, we analyze the system implementation process, including environmental data processing, physical condition monitoring and system services and management, and present the corresponding algorithms. Finally, we carry out an evaluation and analysis from the perspective of insurance and compensation for manufacturing workers in adverse working conditions. The proposed scheme will contribute to the improvement of workplace conditions, realize health monitoring and management, and protect the interests of manufacturing workers.

Keywords

Health monitoring and management Body area networks Cloud computing Industrial wireless networks 

Introduction

In the manufacturing industry, employees working in the adverse working conditions face many adverse factors (e.g., metallic dust and noise), which influence their health condition [1]. Also, serious accidents may occur when some parameters exceed prescribed safety values. For example, the excessive concentration of metallic dust may lead to an explosion. From another perspective, if the physical condition of manufacturing workers is monitored in real time, this could allow the provision of better and timelier health services.

Recently, some emerging technologies, such as Industrial Wireless Networks (IWN) [2, 3, 4], Body Area Networks (BAN) [5, 6, 7, 8], cloud computing [9, 10, 11, 12], smart clothing [13, 14], Big data [15, 16, 17], embedded and pervasive computing [18, 19], and smart sensors [20, 21], have been widely used in health and environmental monitoring systems to reduce safety accidents effectively and accurately assess the health indicators of manufacturing workers in hazardous environments. In addition, Cyber-Physical Systems (CPS) [22, 23, 24] as an enabling technology, are promoting the upgrade of industrial systems and are forming a new era of Industry 4.0 [25, 26]. As research continues, the integration of CPS and cloud technologies is also an inevitable trend which can lead to an improved systemic ability in managing and analyzing big data from industrial environments [27]. In [28], a group-centric intelligent recommender system integrating social, mobile and big data technologies was designed, which can be used for reference of this article.

However, when we design an efficient health monitoring and management system for manufacturing workers in adverse working conditions, some issues and challenges remain unresolved. From the perspective of data types, both environmental parameters and health indicators must be taken into account, which consequently implies a flexible system architecture supporting multi-sensor data fusion. In addition, there is also the urgent problem of integrating sensing architecture with cloud technology to provide optimized services and management. Apart from these factors, one should also take into account related domains integrated with this system (e.g., insurance), and estimate the systemic advantages gained by comparing the proposed scheme with the traditional schemes.

In order to implement health monitoring and management, we designed and analyzed this system from the point of multidisciplinary integration. In this article, our contributions include the following three aspects:
  1. 1)

    We propose a cloud-integrated architecture to monitor and manage environmental parameters and important physiological parameters, as well as analyze functional modules.

     
  2. 2)

    We dissect the system implementation process, including environmental data processing, physical condition monitoring, and optimized service and management.

     
  3. 3)

    We consider the related domains supporting this system and carry out evaluation and analysis from the perspective of insurance schemes and compensation to verify the validity of the proposed system.

     

The rest of this article is organized as follows. Section 2 discusses the cloud-integrated system architecture. Section 3 contains an analysis of the system implementation process. Section 4 provides an evaluation and analysis of the proposed scheme. Finally, Section 5 concludes this article.

System architecture

In [29], a 2G-RFID based E-healthcare system was designed. We propose five-layer system architecture for health monitoring and management of manufacturing workers in adverse working conditions as shown in Fig. 1. It includes 5 layers: the application and service layer, the management layer, the data processing layer, the network and communication layer, and the data collection layer.
Fig. 1

Five-layer architecture for health monitoring and management system

Data collection layer: This layer provides the original data for the entire system, and comprises systems for data sensing, data acquisition, and data processing. The collection layer is based on various protocols (e.g., RFID, Ethernet, Bluetooth, and Radio Frequency). The BAN is responsible for collecting physical condition measurements, such as heart rate, blood pressure, and body temperature. At the same time, environmental conditions, such as metallic dust, noise, and environmental temperature, are monitored through an IWN.

Network and communication layer: This layer provides the basic functionality for achieving end-to-end connectivity of the health monitoring applications. Also, it is responsible for aggregating data from different devices, and packaging them using a specific format so that they can be transmitted. For example, BAN traffic is converted to a sequence file so that it can be processed on a system using Hadoop infrastructure. Routing and MAC protocols can be used in order to realize efficient communication among all kinds of devices.

Data processing layer: This layer is a key unit, which receives the aggregated data (e.g. the aforementioned sequence file) via the communication layer and performs necessary calculations and statistical measurements. In order increase efficiency, big data can be broken down into smaller chunks and each chunk can be processed separately. Then, the small chunks are recombined and stored for future analysis. It is implemented through a parallel processing architecture based on Hadoop with its MapReduce and distributed HDFS file system facilities.

Management and service layers: The management layer is used to deal with various kinds of outcomes, while the service layer provides end-user connectivity for hospitals, emergency departments, ambulances, police stations and insurance companies (e.g., for insurance plans). In addition, they can also follow manufacturing workers’ health progress through continuous analyses of their past medical history and recommend further action based on their bio-signals. Such kinds of applications help doctors familiarize themselves with the current status of the manufacturing workers, and allow them to be aware of their current status and progress.

Figure 2 shows the multi-domain integration approach for health monitoring and management. As we know, implementing a customized insurance service for manufacturing workers based on big-data analytics will protect the interests of both workers and corporations effectively. In this scheme, we need to integrate several domains, including insurance services, medical services, enterprise information (e.g., environmental parameters and physical conditions), the supervision department, domain experts, etc. Effective information acquisition and analysis are the key in this scheme.
Fig. 2

Multi-domain integration for health monitoring and management

System implementation process

In this system, the data from the various types of devices are obtained through BANs and IWNs. The medical condition of a manufacturing worker and the environmental conditions can be monitored by the corresponding health and environmental monitoring systems, and subsequently uploaded to the cloud by means of a smart phone, a Wi-Fi connection, or similar communication systems depending on the manufacturing worker’s location (e.g., work environment, hospital or outdoor environment).

Environment data processing

In adverse working conditions, the focus should lie on key environmental parameters, such as metallic dust, noise, and temperature. As we know, metallic dust concentration beyond a certain level may cause an explosion. Also, both the excessive noise and excessively high or low temperatures may affect the health of manufacturing workers. With the support of emerging technologies, environmental data processing can help us assess the working environment and prevent accidents.

An indicative algorithm was designed for environmental data processing, which uses the relevant parameters shown in Table 1. Read_par() is responsible for gathering sensor data from different channels. Before Send_data(), all the data will be preprocessed by the function Data_fusion(). Assessment() is the implementation of an expert system, such as a Machine Learning (ML) classification algorithm, complex medical problem detection algorithms, or statistical calculations based on various parameters. Warning() alerts the application layer service to perform quick actions based on received sensor data flagged as important or urgent or based on the analysis results generated by processing servers. Algorithm 1 is the pseudocode for environmental data processing.
Table 1

Parameters and functions used in Algorithm 1

Symbol

Description

I_Channel_ID

Channel for parameter acquisition

I_Alarm_msg

Warning message

I_Read_par ( )

Sensor data reading

I_Data_fusion ( )

Data fusion before transmission

I_Send_data ( )

Data transmission to the cloud

I_Assessment ( )

Data evaluation process using Hadoop

I_Warning ( )

Warning message sent to users

I_Record ( )

Data are stored in the database

During environmental data processing, some related problems are taken into account. First, the system’s administrative authority is considered. For example, when some accidents happen, who should receive the alerting message? Secondly, information security cannot be ignored. Finally, this system involves many domains, which implies that the design should be based on synergism.

Algorithm 1 Pseudocode of environment data processing

1: Initialization;

2: I_Channel_ID ← metallic_dust, noise, environment_temp;

3: Calculation of I_Alarm_msg

4:{ if (timer == set_point)

5: I_Read_par(Channel_ID);

6: I_Data_fusion();

7: I_Send_data();

8: end if

9: I_Result ← I_Assessment(); //Hadoop

10: if(I_Result ≤ threshold)

11: I_Record();

12: else

13: I_Record();

14: I_Alarm_msg ← I_Warning();

15: end if

15: ReturnI_Alarm_msg;

16:}

Physical condition monitoring

The physical condition parameters of manufacturing workers mainly include the workers’ heart rate, blood pressure and body temperature. Since these parameters directly represent the workers’ physical condition, we can adopt some effective measures to protect manufacturing workers if we can obtain these parameters in a timely manner and provide effective assessment. In this system, the sensors obtain the raw data and save them in the memory of the sensor boards. With the support of BANs, mobile phones can periodically read the raw data and transmit them to the cloud to provide the foundation for further analysis and evaluation.

Table 2 shows the parameters and functions used in Algorithm 2. The two main parameters, B_Channel_ID and B_Body_msg, represent the parameter acquisition channel and warning messages, respectively. B_Sensor_record() is responsible for recording raw data, while B_Read_par() obtains sensor data through the BAN and stores them in the phone through the function B_Save_data(). Subsequently, the function B_Data_preproc() completes the preprocessing of the data before the function B_Send_data() is used to transmit the data to cloud. B_analyze_data() represents the Hadoop-based data evaluation process, while the result is saved into the database using the B_Record() function. Of course, if the result obtained is not within certain safety margins, an event is triggered to inform the affected users. At the same time, the processing plan (e.g., insurance services) is initiated. Algorithm 2 provides the pseudocode of the physical condition monitoring.
Table 2

Parameters and functions used in Algorithm 2

Symbol

Description

B_Channel_ID

Channel for parameter acquisition

B_Body_msg

Warning message

B_Sensor_record()

Raw data recording

B_Read_par()

App. on the phone obtains sensor data

B_Save_data()

Storage of data on the phone

B_Data_preproc()

Data processing before transmission

B_Send_data()

Data transmission to the cloud

B_analyze_data()

Data evaluation using Hadoop

B_Start_proc()

Processing plan initiation

B_Record()

Data storage in the database

Decision servers are equipped with intelligent medical expert systems, ML classifiers, and other complicated medical condition detection algorithms for further analysis and decision-making. The results received from the processing unit are analyzed by complex medical expert systems, machine ML classifiers, etc., using the medical history of the manufacturing worker. The concrete details of the medical expert systems are beyond the scope of the article due to the extent of the subject. Usually, decision servers machine use learning classifiers like REPTree, which are more efficient and accurate for various normal disease detections [12, 15].

Algorithm 2 Pseudocode of physical condition monitoring

1: Initialization;

2: B_Channel_ID ← heart_rate, blood_pressure, body_temp;

3: B_Sensor_record();

4: Calculation of B_Body_msg

5: { if (timer == set_point)

6: B_Read_par(B_Channel_ID);

7: B_Save_data();

8: B_Data_preproc();

9: B_Send_data();

10: end if

11: B_result ← B_analyze_data(); //Hadoop

12: if(B_resultnormal)

13: B_Body_msg ← B_Start_proc();

14: else

15: B_Record();

16: end if

17: ReturnB_Body_msg;

18:}

Optimized service and management

This system facilitates the exchange of information among many domains, such as physical health, factory environment, hospitals, insurance services, and government agencies. From the perspective of technologies, the monitoring of the workers’ physical health and the environmental conditions is based on BANs and IWNs, respectively. In order to further analyze the raw data and formulate an intelligent decision, the cloud platform is the foundation to carry out computing and storage. During the analysis, Hadoop is adopted to process the big data from the manufacturing environment and the workers. Important results will be forwarded to the corresponding users. For example, if a manufacturing worker suddenly becomes ill and has to be hospitalized, a message should be sent to family members and nearby hospitals.

In adverse working conditions, the environmental parameters (e.g., metallic dust concentration, noise, and temperature) directly affect the health conditions of manufacturing workers. If one can effectively monitor and assess these parameters, the environmental safety of working can be ensured and accidents can be avoided. Also, these parameters are saved in the cloud, and this can form the basis for demonstrating the safe working conditions of an enterprise. Of course, according to the assessment result, the decision maker may implement a plan for further improvement of working conditions.

Ensuring manufacturing workers’ health is very important and health monitoring and management based on fundamental physiological parameters (e.g., heart rate, blood pressure, and body temperature) is a direct way to achieve this end result. In this application, smart phones form the main means of data preprocessing before transmission to the cloud. The medical history data of each manufacturing worker contained in the cloud are essential for assessing their health level, and can be used to formulate health insurance programmes. Here, a practical issue is that the manufacturing worker will have to carry some sensors and a smartphone, and this will probably affect their performance. In addition, the real-time performance of this system must also be considered.

This article contributes to the synergy and integration among multi-disciplinary fields. To justify the relevant points more concretely, we consider the aspects, such as insurance considerations, environmental monitoring parameters, and health conditions. The sensing, wireless, and cloud computing technologies are the foundation for realizing this system. The analyses and decisions taking place in the cloud are the key to ensuring quality of services. Figure 3 shows the flowchart of optimized service provision and management.
Fig. 3

Flowchart of optimized service provision and management

Although the basic concepts for the implementation of this scheme have been clearly formulated, there are still some issues and challenges to be faced as follows:
  1. 1)

    The methods of big data analysis and decision making directly affect the system’s overall performance. Also, the data format and efficient information interaction should be carefully considered.

     
  2. 2)

    The integration of the insurance plan with health condition information directly determines the economic benefit. Whether or not a manufacturing worker should be allowed to finish an inspection or be sent for a diagnosis will depend on the evaluation results obtained from the proposed system.

     
  3. 3)

    This system involves much data from different domains. Permission settings, privacy and security are key aspects to be considered.

     
  4. 4)

    Data fusion before transmission to the cloud is very important in order to improve transmission efficiency.

     

Evaluation and analysis

Based on the previous analysis, we can carry out an evaluation and analysis of this system on the basis of emerging technologies and using a statistical approach. Table 3 shows the supporting technologies for evaluation and analysis of the proposed health monitoring and management system. As a whole, these technologies can be divided into two categories. The first category includes data collection and transmission, which includes all kinds of sensors, near-field-communications (e.g., WSNs, BANs, and IWNs), and long distance wireless communications (e.g., 3G or 4G). The other category comprises the data processing platform and tools. The operating system and server management tools are XenServer + XenCenter and CentOS, and constitute the basic platform. We chose Hadoop and Hbase as the distributed computing framework and database, respectively. Other auxiliary tools include AVS, NLPIR, and HANA.
Table 3

Supporting technologies for evaluation and analysis

Technologies

Description

WSN

Wireless sensor networks

BAN

Body area network

IWN

Industrial wireless network

Sensors

Sensing environment or body

XenServer + XenCenter

Server management

CentOS

Node operating system

Hadoop

Distributed computing framework

Hbase

Database

AVS

Visualization tools

NLPIR

Text semantic analysis

HANA

Memory computing technology

Android

Mobile application development

In order to simplify the analysis, we assume some initial parameter values. Table 4 shows the parameters used for the evaluation and analysis process. We monitored three kinds of parameters (heart rate, blood pressure, and body temperature) for 50 manufacturing workers, and three kinds of environmental parameters (metallic dust, noise, and environment temperature). We also assume that abnormal probability for body and the probability of an accident happening within a given year are fixed values. The medical cost of each case and accident cost are taken from statistical averages. In addition, we also assume that the cost of establishing the environmental monitoring system amounts to a total of thirty thousand yuan.
Table 4

Initial parameters for evaluation and analysis

Parameters

Value

Description

Num_peo

50

Number of people

Body_par

3

Three types of parameters Number of physical condition indicators

Env_par

3

Number of environmental conditions’ indicators

Body_pro

3 %

Abnormal probability for body

Env_pro

0.5 %

Accident probability of year

Insur_char

60¥

Insurance charge of year

Med_cost

10,000¥

Medical cost of each case

Accid_cost

100,000¥

Accident cost of each case

According to the initial parameters for evaluation and analysis in Table 4, we can calculate the medical and accident costs based on the statistics as shown in Fig. 4. With the support of new technologies, we may assess workers’ physical and environmental conditions in real time, which may give some references and suggestions for implementing health insurance programmes and avoiding accidents. Of course, we also know that the real conditions modelled by this system are stochastic. For example, sickness usually follows the logistic regression. As we can see from Fig. 4, the health monitoring and management system in adverse working conditions can protect the interests of manufacturing workers and corporations.
Fig. 4

Medical and accident costs according to statistical analysis

Based on the above analysis, we only give the basic results under the conditions of the hypothesis. In real applications, we should consider many additional issues, such as real-time performance, information privacy, and system information interaction. However, since this system involves many domains, it is challenging to effectively integrate and monitor all the elements.

Conclusions

With the support of some emerging technologies, it is possible for us to effectively reduce workplace accidents, and evaluate the health indicators of manufacturing workers in poor environmental conditions. In this article, we focus on three aspects. First, we propose a cloud-integrated architecture to monitor and evaluate important environmental and physiological parameters, and analyze the functional modules. Then, we dissect the system implementation process, including environment data processing, physical condition monitoring, and optimized service provision and management. Finally, we consider the necessary cross-domain interactions for supporting this system, and carry out an evaluation and analysis from the perspective of insurance and compensation; this analysis verified the validity of the proposed system. In future works, we will implement specific verification and assessment methods for real applications.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (Nos. 61572220, and 61262013), the Fundamental Research Funds for the Central Universities (No. 2015ZZ079), the Water Resource Science and Technology Innovation Program of Guangdong Province (No. 2016-18), the Natural Science Foundation of Guangdong Province, China (Nos. 2016 A030313734 and 2016 A030313735), and the Quality Project of Guangdong Province Office of Education (No. 2016-135).

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  • Xiaoya Xu
    • 1
  • Miao Zhong
    • 2
  • Jiafu Wan
    • 3
  • Minglun Yi
    • 4
  • Tiancheng Gao
    • 5
  1. 1.Guangdong Mechanical & Electrical CollegeGuangzhouChina
  2. 2.Guangdong University of EducationGuangzhouChina
  3. 3.School of Mechanical & Automotive EngineeringSouth China University of TechnologyGuangzhouChina
  4. 4.School of Electrical Engineering and AutomationJiangxi University of Science and TechnologyGanzhouChina
  5. 5.North Information Control Group Company LimitedNanjingChina

Personalised recommendations