INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1073
Stratification of Medical Equipment Using Clustering Algorithm and
Optimized Maintenance Scheduling
Srilekha S
Centre for Distance Education, Anna University, Chennai.
DOI: https://doi.org/10.51244/IJRSI.2025.1210000096
Received: 06 October 2025; Accepted: 14 October 2025; Published: 05 November 2025
ABSTRACT
Medical equipment maintenance in healthcare facilities requires strategic prioritization to optimize resource
allocation and ensure patient safety. This study presents a novel approach for stratifying medical equipment
using K-means clustering algorithm combined with optimized maintenance scheduling. A comprehensive
dataset of 1,973 medical equipment from X Hospital, Chennai, was analyzed using features including purchase
cost, downtime, usage patterns, and preventive maintenance costs. The clustering algorithm successfully
stratified equipment into three priority categories: High Priority (658 equipment, 33.4%), Medium Priority (657
equipment, 33.3%), and Low Priority (658 equipment, 33.4%). The silhouette score of 0.154 indicates reasonable
clustering validity. Optimized maintenance scheduling based on priority stratification resulted in estimated
annual cost savings of Rs. 1,580,337 (4.26% reduction) and downtime reduction of 20,207 days (17.1%
improvement). High-priority equipment received monthly preventive maintenance intervals (30 days), medium-
priority equipment received bi-monthly intervals (60 days), and low-priority equipment received quarterly
intervals (90 days). The implementation requires 14,470 annual PM activities, 59,194 inspections, and 4,604
calibrations, totaling 79,584 maintenance hours annually. The study demonstrates that data-driven equipment
stratification can significantly improve maintenance efficiency and reduce operational costs in healthcare
settings.
Keywords: Medical equipment, clustering algorithm, maintenance scheduling, healthcare management,
predictive maintenance
INTRODUCTION
Medical equipment forms the backbone of modern healthcare delivery, encompassing thousands of devices
ranging from simple diagnostic tools to sophisticated life-support systems. Healthcare facilities typically manage
diverse equipment portfolios with varying operational requirements, maintenance needs, and patient safety
implications. Traditional maintenance strategies have historically relied on manufacturer-recommended
schedules or reactive repairs following equipment failures, failing to account for varying criticality levels, usage
patterns, and impact on patient care.
When diverse equipment types are managed uniformly, the result is often suboptimal resource allocation,
increased operational costs, and potential safety risks. A ventilator used in an intensive care unit requires
fundamentally different maintenance attention compared to a wheelchair or routine blood pressure monitor, yet
many healthcare facilities apply similar maintenance protocols across all equipment categories.
The stratification of medical equipment and optimization of maintenance scheduling is increasingly recognized
as a crucial strategy for enhancing reliability and operational efficiency within healthcare settings. Recent studies
have underscored the importance of data-driven approaches and machine learning models in enabling effective
maintenance management.
This research addresses the critical need for intelligent equipment grouping based on usage patterns and
importance, enabling more focused and timely maintenance through clustering algorithms and optimized
scheduling strategies.
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1074
LITERATURE REVIEW
The application of clustering algorithms in healthcare equipment management has gained significant attention
in recent years. Boppana (2023) emphasized the role of data analytics in predictive maintenance for healthcare
equipment, highlighting how real-time data and advanced algorithms facilitate early fault detection and
maximize equipment uptime [4].
Roy Chowdhuri et al. (2023) developed structured prioritization techniques aimed at preventing unexpected
device failures, thereby supporting patient safety and continuity of care [5]. Similarly, Akpan and Anyi-
Akparanta (2024) introduced hospital-based reliability models employing statistical approaches to forecast
equipment failure probabilities, which guide maintenance scheduling and resource allocation more effectively
[6].
Zamzam et al. (2021) showcased the application of unsupervised machine learning algorithms—particularly K-
means clustering—to stratify medical equipment according to preventive, corrective, or replacement
maintenance priorities. Such data-driven stratification aids clinical engineers by making workload prioritization
more objective and efficient [7] .
The World Health Organization (2025) advocates for integrated inventory and maintenance management
systems that enable accurate device stratification and scheduling, ensuring the safety and availability of medical
devices in healthcare facilities [8].
Alahmadi et al. (2025) advanced the development of predictive maintenance models integrating machine
learning and optimization algorithms, marking a transition from reactive to reliability-centered, data-informed
maintenance strategies [9] . Ma et al. (2023) applied artificial intelligence to predict the remaining useful life
and potential failure of medical equipment, thereby supporting proactively planned maintenance interventions
[10].
METHODOLOGY
Data Collection and Dataset
The study utilized a comprehensive dataset of 1,973 medical equipment records from Kilpauk Hospital, Chennai.
The dataset included equipment spanning multiple categories (A, B1, B2, C) with varying criticality levels and
maintenance requirements.
Feature Selection
Four primary features were selected for clustering analysis:
Purchase Cost (Rs.): Equipment acquisition value
Downtime per Year (Days): Annual equipment unavailability
Usage per Week (Hours): Weekly operational hours
Preventive Maintenance Cost per Year (Rs.): Annual maintenance expenses
Data Preprocessing
Data preprocessing involved standardization using StandardScaler to ensure comparable feature scales across all
variables. Missing values were minimal (less than 2%) and were handled through appropriate imputation
techniques.
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1075
Clustering Algorithm Implementation
K-means Algorithm: The study implemented K-means clustering with the following parameters:
Number of clusters: 3 (determined through silhouette analysis)
Random state: 42 for reproducibility
Maximum iterations: 300 Initialization method: k-means++
Criticality Score Calculation: A composite maintenance criticality score was calculated using weighted
features:
Optimization Framework
Priority-based maintenance intervals were established:
High Priority: 30-day PM intervals, 7-day inspections, 90-day calibrations
Medium Priority: 60-day PM intervals, 14-day inspections, 180-day calibrations
Low Priority: 90-day PM intervals, 30-day inspections, 365-day calibrations
RESULTS AND DISCUSSION
Clustering Results
The K-means algorithm successfully stratified the 1,973 medical equipment into three distinct priority categories
with nearly equal distribution, as shown in Figure 1.
Figure 1: Equipment Distribution by Maintenance Priority
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1076
The silhouette score of 0.154 indicates reasonable clustering validity, suggesting meaningful separation between
equipment groups based on maintenance characteristics.
Cluster Characteristics Analysis
Table I presents the detailed characteristics of each equipment cluster, demonstrating distinct maintenance
profiles across priority categories.
Table I: Equipment Cluster Characteristics
High Priority Equipment demonstrated the highest maintenance criticality with average downtime of 103.23
days/year. Primary equipment types included multiparameter monitors (164 units), ventilators (141 units), and
syringe infusion pumps (137 units).
Medium Priority Equipment showed moderate maintenance requirements with 49.31 days average downtime
per year. This category included 242 multiparameter monitors, 87 pulse oximeters, and 83 ventilators.
Low Priority Equipment exhibited lower maintenance intensity with 26.80 days average annual downtime.
Common equipment included 115 infusion pumps, 74 syringe infusion pumps, and 61 defibrillators.
Optimized Maintenance Intervals
Table II presents the optimized maintenance scheduling parameters for each priority category. The differentiated
approach ensures that critical equipment receives more frequent attention while optimizing resource allocation.
Priority Category PM Interval Inspection Interval Calibration Interval Annual PM Frequency
High Priority 30 days 7 days 90 days 12 times/year
Medium Priority 60 days 14 days 180 days 6 times/year
Low Priority 90 days 30 days 365 days 4 times/year
Table II: Optimized Maintenance Scheduling Parameters
Figure 2 illustrates the differentiated maintenance intervals across priority categories, demonstrating the strategic
allocation of maintenance resources.
Priority
Category
Count
Mean Cost
(Rs)
Mean
Downtime
(Days)
Mean Usage
(Hrs/Week)
Mean
PMCost (Rs)
Low Priority 658 200,351 26.80 90.04 18,032
Medium
Priority
657 195,237 49.31 139.21 17,571
High Priority 658 231,314 103.23 133.45 20,818
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1077
Figure 2: Optimized Maintenance Intervals by Priority Category
Maintenance Activities Distribution
The optimized scheduling framework generates specific annual maintenance activities for each priority category:
High Priority: 7,896 PM activities, 34,216 inspections, 2,632 calibrations
Medium Priority: 3,942 PM activities, 17,082 inspections, 1,314 calibrations
Low Priority: 2,632 PM activities, 7,896 inspections, 658 calibrations
Figure 3 shows the distribution of annual maintenance activities across priority categories.
Figure 3: Annual Maintenance Activities Distribution by Priority
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1078
The total annual maintenance workload comprises:
Total PM Activities: 14,470
Total Inspections: 59,194
Total Calibrations: 4,604
Total Maintenance Hours: 79,584 hours (9,948 person-days)
Required Staff: 39 full-time maintenance technicians
Performance Comparison
Table III presents a comprehensive comparison between the current system and the optimized system,
highlighting significant improvements in cost and efficiency.
Metric Current System Optimized System
Total Equipment 1,973 1,973
Annual PM Cost (Rs) 37,107,565 35,527,228
Annual Downtime (Days) 117,954 97,747
PM Activities/Year Variable 14,470
Maintenance Hours/Year Unoptimized 79,584
Cost Reduction (%) - 4.26%
Downtime Reduction (%) - 17.1%
Table III: Performance Comparison - Current Vs Optimized System
Figure 4 visualizes the comparison between current and optimized maintenance performance.
Figure 4: Current vs Optimized Maintenance Performance
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1079
Downtime Reduction Analysis
Table IV presents detailed downtime reduction results by priority category.
Priority
Category
Current Downtime
(Days)
Optimized Downtime
(Days)
Reduction
(Days)
Reduction (%)
High Priority 67,925 54,340 13,585 20.0%
Medium
Priority
32,395 27,536 4,859 15.0%
Low Priority 17,634 15,871 1,763 10.0%
Total 117,954 97,747 20,207 17.1%
Table IV: Downtime Reduction Analysis by Priority
Figure 5 illustrates the downtime reduction achieved across priority categories.
Figure 5: Downtime Reduction Analysis by Priority Category
The high-priority category achieved the largest absolute downtime reduction of 13,585 days (20%),
demonstrating the effectiveness of intensive maintenance scheduling for critical equipment. Medium and low-
priority categories achieved 15% and 10% reductions respectively, balancing resource efficiency with
maintenance effectiveness.
Equipment Type Distribution
Table V shows the top equipment types identified in each priority category, validating the clinical relevance of
the stratification approach.
Priority Rank Equipment Type Count
High Priority 1 Multiparameter Monitor 164
2 Ventilator 141
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1080
3 Syringe Infusion Pump 137
4 Infusion Pump 104
5 Syringe Pump 64
Medium Priority 1 Multiparameter Monitor 242
2 Pulse Oximeter 87
3 Ventilator 83
4 Infusion Pump 72
5 Syringe Infusion Pump 71
Low Priority 1 Infusion Pump 115
2 Syringe Infusion Pump 74
3 Defibrillator 61
4 Electrocardiograph 57
5 Multiparameter Monitor 49
Table V: Top Equipment Types by Priority Category
The distribution aligns with clinical expectations, with life-support equipment (ventilators, infusion pumps)
appropriately classified in high-priority categories, while diagnostic equipment (ECG, pulse oximeters) appears
across multiple categories based on usage patterns and criticality.
Clustering Visualization
Figure 6 demonstrates the relationship between equipment downtime and maintenance costs, with clear
separation between priority clusters.
Figure 6: Equipment Clustering by Downtime and Maintenance Cost
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1081
The scatter plot reveals distinct clustering patterns, with high-priority equipment generally exhibiting higher
downtime values, while maintenance costs vary across priority levels based on equipment complexity and usage
intensity.
CONCLUSION
This study successfully demonstrates the effectiveness of clustering algorithms in stratifying medical equipment
for optimized maintenance scheduling. The K-means clustering approach, combined with criticality scoring,
provides a robust framework for equipment prioritization that addresses the heterogeneous nature of medical
equipment maintenance requirements.
The achieved results represent significant operational enhancements:
Cost Reduction: Rs. 1,580,337 annual savings (4.26% reduction)
Downtime Reduction: 20,207 days annual improvement (17.1% reduction)
Structured Scheduling: 14,470 PM activities, 59,194 inspections, 4,604 calibrations
Resource Planning: Clear requirements of 79,584 annual maintenance hours
The balanced distribution across priority categories (33.3-33.4% each) ensures practical implementation while
maintaining clinical relevance. High- priority equipment, including ventilators and multiparameter monitors,
receives appropriate intensive maintenance (30-day intervals), while lower- priority equipment follows less
frequent but adequate schedules.
The proposed framework offers healthcare institutions a data-driven approach to maintenance management that
improves equipment reliability, reduces operational costs, and enhances patient safety. The methodology is
scalable and can be adapted across different healthcare settings and equipment portfolios.
Future Research Directions include:
Integration of real-time sensor data for dynamic priority adjustment
Development of failure prediction models using machine learning
Expansion to include predictive maintenance algorithms
Cost-benefit analysis across multiple healthcare institutions
Implementation of IoT-based monitoring systems
The stratification methodology presented provides a foundation for evidence-based maintenance management
that contributes to more efficient and effective medical equipment management practices in healthcare facilities.
REFERENCES
1. S. S. Srilekha, "Stratification of Medical Equipment Using Clustering Algorithm and Optimized
Maintenance Scheduling - A First Review Report," Anna University, Chennai, 2025.
2. V. Boppana, "Data analytics for predictive maintenance in healthcare equipment," EPH - International
Journal of Business & Management Science, vol. 9, no. 2, pp. 26-87, 2023.
3. I. Roy Chowdhuri, A. Kar, and A.K. Pal, "A New Approach for Effective Reliability Management of
Biomedical Equipment," European Journal of Theoretical and Applied Sciences, vol. 1, no. 5, pp. 281-
293, 2023.
4. N.P. Akpan and E.R. Anyi-Akparanta, "Reliability Model of Medical Equipment in University of Port
Teaching Hospital," Earthline Journal of Mathematical Sciences, vol. 14, no. 3, pp. 459-475, 2024.
INTERNATIONAL JOURNAL OF RESEARCH AND SCIENTIFIC INNOVATION (IJRSI)
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Page 1082
5. A.H. Zamzam et al., "A systematic review of medical equipment reliability assessment in improving the
quality of healthcare services," Frontiers in Public Health, vol. 9, pp. 1-12, 2021.
6. World Health Organization, "Inventory and maintenance management information system for medical
devices," WHO Medical Device Technical Series, Geneva: World Health Organization, pp. i–125, 2025.
7. A.H. Zamzam et al., "Prioritisation Assessment and Robust Predictive System for Medical Equipment:
A Comprehensive Strategic Maintenance Management," Frontiers in Public Health, vol. 9, Article
782203, pp. 1-19, 2021.
8. K.M. Alahmadi, E.R.I. Mahmoud, and F. Imaduddin, "Model Development to Improve the Predictive
Maintenance Reliability of Medical Devices," IAPGOŚ, vol. 1, pp. 117-124, 2025.
9. H. Ma, C. Xu, and J. Yang, "Design of Fine Life Cycle Prediction System for Failure of Medical
Equipment," Journal of Artificial Intelligence and Technology, vol. 3, no. 2, pp. 39–45, 2023.
10. A. Shamayleh, M. Awad, and J. Farhat, "IoT Based Predictive Maintenance Management of Medical
Equipment," Journal of Medical Systems, vol. 44, Article 72, pp. 1–12, 2020.