Table 3.

Existing literature on medical devices reliability and research gap.

Topic	Technique/area of concern	Geographic area	Medical device	Articles/year	Outcomes	Research gap/future work
Risk management (prioritization)	Fuzzy AHP	Portugal	Five selected medical devices	[42]/2020	Priority for renewing medical device categorized to low, medium, high and urgent for replacement	Only selected medical devices are included, and the most significant criteria are not specified.
						Limited no. of age in service
	RCM	Malaysia	Medical device	[43]/2019	Breakdown factors are divided into three which is maintenance services type, environmental and human factors	RCM method is widely used but requires sufficient data to complete the process
	Mathematical model	Mexico	16 selected medical devices	[44]/2020	The result determines the annual number and priority for PM. Type of equipment (highest priority), location (lowest priority)	The model applies to selected 16 medical devices
		South Africa	Infusion pump and ventilator	[20]/2013	The findings conclude age does not affect the survival of equipment due to the limited number of 5 years in service	Limited data for ventilators and only five years in age.
						Model not possible to be analysed to other devices due to insufficient data
		United States, France	Medical device	[14]/2010	Biomedical equipment is classified into the high, medium, and low category	Current maintenance strategies are effective but lack the evidence of being efficient
	AHP	Jakarta	Medical device	[12]/2019	The highest maintenance priority is an excimer laser, followed by a retinal laser and others	The most significant parameters which influence the result is not specified
		Canada	Medical device	[45]/2011	Maintenance is prioritized with scores. Higher score for high priority in maintenance management program	A higher score requires further investigation
	Logistic regression predictor model	Romania	Three selected medical devices	[46]/2017	Maintenance intervals are prioritized and developed based on risk group from low to high	No standard exists for assessing risk, and the tool uses current practices to establish a baseline
	AHP, TOPSIS, and MILP	Tunisia, Africa	Medical device	[47]/2017	The maintenance strategies framework is developed and divided into time-based, condition-based and corrective maintenance	The framework shall be enhanced to a mathematical model
	PVST	Istanbul, Turkey	16 selected medical devices	[6]/2016	The preventive maintenance schedule is developed for older technology and predictive maintenance for newer technology	A future study is required to investigate failures of other medical devices excluded from this study.
	QFD	Italy	Four selected medical devices	[48],	A comprehensive framework for PM priority is developed. The most important criteria are function, maintenance requirement, and others	Data was collected in 2012. The framework is tested only once during scheduled PM.
				[49]/2015
Risk management (failure and risk analysis)	FFMEA	India	Ventilator	[50]/2020	Nine failure modes for ventilators are ranked based on the risk criteria from remote, low to very high	Application of fuzzy FMEA limited to ventilator
		Budapest, Hungary	Medical device	[51]/2016	Comparison between FMEA and FFMEA technique concludes FMEA is more accurate by involving different expert impacts weights	Difficulties in collecting each expert opinion for each risk as to the number of risks increases
		Canada	Five selected medical devices	[52]/2015	Framework for prioritization and budget allocation for maintenance are developed. A maintenance strategy is proposed from low to very high priority	Development of a risk-based maintenance software based on the suggested comprehensive framework
	Six sigma, Pareto analysis	India	Ventilator	[53]/2020	Common failures are identified, which are flow sensor, expiratory valve, calibration, battery, display and oxygen sensor	Automated real-time proactive RCA shall be enhanced to prevent failures
	RCM and FMEA	United Arab Emirates	Four selected medical devices	[54]/2020	Results examine the relationship between the current practice of PM, failure mode, and RCM action is identified based on failure modes	Failure data for only 1 year.
						Lack of maintenance cost data and limitation to run RCM pilot project in the hospital
	FMEA, Pareto analysis and 5 whys	Kenya, Belgium	Three selected medical devices	[55]/2018	Daily, weekly, monthly tasks and maintenance protocols are developed based on the failures identified	Only focuses on cobalt-60 radiotherapy machines and limitations to evaluate the effectiveness of proposed strategies on operation and maintenance
	FMEA	Italy	Medical device	[56]/2015	One leading company in the development of medical devices is selected, and the process/risk connected to the design of new devices are evaluated	Technique shall be applied to other leading companies and develop a more manageable approach to overcome FMEA limitation
		Sierra Leonne, Africa	Anesthesia machine	[57]/2014	Five failures mode are identified, which are resource availability, environmental, staff knowledge and attitudes, workload, and staffing	The sample size is small, with only two hospitals involved, and it is not convincing whether the findings are applicable to other settings
		China	Medical device	[58]/2014	Potential failures in human reliability of medical devices are evaluated with numerical values of risk factors	Propose to apply the model to different medical device design
Performance prediction for medical device using machine learning	Machine learning (artificial neural network and fuzzy logic classifier)	Sarajevo, Bosnia, and Herzegovina	Infant incubator	[29]/2020	Performance is predicted, and decision tree has the best properties compared to the other four algorithms with 98.5% accuracy based on performance output error	Model is applicable for infant incubators only with two years dataset period
	Machine learning (artificial neural network)	Sarajevo, Bosnia, and Herzegovina	Infusion and perfusor pumps	[31]/2020	Feedforward neural network with ten neurons in a single hidden layer has an accuracy of 98.06% for perfusion pumps, 98.83% for infusion pumps, and 98.41% for both	Research shall be extended by introducing new parameters such as maintenance history and spare parts replacement to enhance accuracy
Performance prediction for medical device using machine learning–cont'd	Machine learning	Bosnia and Herzegovina	Defibrillator	[32]/2019	Performance is predicted, and random forest has the best properties compared to the other four algorithms with 100% accuracy	Model is applicable for defibrillator only with three years dataset period
Medical devices management system (MDMS) (marketing strategies)	AHP (questionnaire)	Iran	Medical device	[34]/2019	Research on marketing strategies concludes most essential barriers are a managerial and strategic barrier	Fuzzy AHP technique shall be applied to examine the compatibility with human verbal and vague expressions
MDMS (management system)	Qualitative approach (interview)	Iran	Medical device	[4]/2019	Factors influencing medical device management systems are categorized into seven themes (resources preventive maintenance, design, implementation, etc.), with 19 subthemes	The themes subjected for further research in Iran or other countries to improve quality
MDMS (service quality)	AHP	Iran	Medical device	[36]/2019	4 Iranian public hospitals are ranked based on four criteria to evaluate hospital service quality	More hospital selection would provide a better benchmark
MDMS (service quality)	Qualitative approach (questionnaire)	Ghana, West Africa	Medical device	[35]/2019	Adequacy of healthcare resources is the most decisive factor compared to the other four service quality factors on patient satisfaction	Shall be enhanced to the district or regional hospital instead of a teaching hospital
MDMS (management system)	Literature review	Iran	Medical device	[8]/2018	Eighty-nine factors affected medical equipment maintenance management: Resources, education, service, quality, inspection, etc.	Some factors overlapped with each other
MDMS (maintenance strategies)	MCDM	Morocco	Medical device	[2]/2018	Maintenance strategies conclude risk (29%) as an essential criterion, followed by equipment function (14%).	Data collection is lacking and investigating external contributors such as heavy use and misuse and environmental factors on hidden failure event
MDMS (replacement plan)	—	Lebanon	35 selected medical devices	[37]/2016	A replacement plan is proposed with ranked criteria and sub-criteria depending on the urgency	Integration between hardware and software using Internet shall be executed to generate data on lifespan
MDMS (documentation)	—	Bangladesh	Ventilator	[59]/2015	Risk factor reduction and standard operating procedure (SOP) is developed. Concludes lack of adequately educated, and trained clinical engineers to be solved	Service contract with vendors for maintenance shall be developed
MDMS (utilization and human resources)	Qualitative approach (questionnaire)	India	Diagnostic medical device	[10]/2015	23% of the devices are underutilised	The sample size was small and limited to diagnostic devices in the histopathology lab in 2012
MDMS (quality assurance)	—	Bucharest, Romania	Radiant warmer, and infusion pump	[38]/2013	Risk and score for both devices are addressed with five different criteria as guidance in managing quality assurance program	Limited to only two types of medical devices. Maintenance software in the database shall be developed
MDMS (management system/Software)	—	Jordan	Medical device	[60]/2012	Presents a software system (EQUI-MEDCOMP) using microsoft visual basic (version 6) designed to improve maintenance management	Parameters used in the dataset are limited to 5 factors; other factors shall also be considered
MDMS (quality control)	—	China	9 selected medical devices	[39]/2010	A six-dimension risk model is proposed, and a quality control system is established	Quality control shall be enhanced to more types of medical devices