Page 1179
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Predictive Maintenance Analytics and Fleet Downtime Reduction in
U.S. Car Rental Operations
Salami Abdul Mohammed
College of Business, Westcliff University, Irvine, California, USA
DOI:
https://doi.org/10.51583/IJLTEMAS.2026.150500091
Received: 12 May 2026; Accepted: 16 May 2026; Published: 03 June 2026
ABSTRACT
Fleet predictive maintenance analytics are reshaping vehicle maintenance management in U.S. car rental
operations, enabling operators to predict component failures before they cause downtime. The independent
variable in this analysis is predictive maintenance analytics adoption, defined as the deployment of IoT sensors,
machine learning models, real-time alert systems, and explainable AI output interfaces across rental vehicle
fleets. The dependent variable is fleet downtime reduction outcomes, operationalized as reductions in unplanned
vehicle downtime, maintenance costs, safety incidents, and customer service failures attributable to vehicle
mechanical failures. Drawing on Human Capital Theory and the Technology-Organization-Environment
framework, the paper conducts a systematic narrative literature review across machine learning, vehicle
maintenance, fleet management, cybersecurity, and operations management. It reviews ML and explainable AI
techniques for vehicle fleet maintenance prediction, maps predictive maintenance demands by vehicle system,
evaluates adoption barriers including cybersecurity and data privacy risks using the TOE framework, and
proposes a four-stage implementation framework with verified cost ranges and ROI measurement approaches.
The paper's principal finding is that barriers to effective predictive maintenance are concentrated in the
organizational dimension: sensor infrastructure is increasingly available and ML models are deployable, but the
capability of operations staff to interpret, evaluate, and act on predictive alerts is not being developed
systematically. Braking and tire systems carry safety and legal compliance dimensions that make human
oversight a regulatory requirement as well as an operational one. The paper contributes an integrated predictive
maintenance framework specific to car rental fleet operations, the first vehicle system risk matrix calibrated to
rental fleet oversight demands, a TOE-grounded barrier analysis incorporating cybersecurity risks, a four-stage
implementation model with ROI measurement approach, and a proposed empirical validation design for primary
data confirmation.
Keywords: Predictive maintenance; Fleet management; Car rental operations; Machine learning; Human
Capital Theory
INTRODUCTION
Fleet maintenance management is one of the most operationally consequential and cost-intensive functions in
car rental operations. The physical condition of every vehicle in a rental fleet directly determines customer safety,
service reliability, vehicle availability, and maintenance cost. These four performance dimensions together
distinguish profitable operators from unprofitable ones over time. The emergence of AI-driven predictive
maintenance analytics, enabled by the integration of IoT sensors, telematics platforms, machine learning models,
and explainable AI output systems, has created a fundamentally new set of capabilities for car rental fleet
management: the ability to predict vehicle failures before they occur, schedule maintenance interventions at
operationally optimal times, and prevent the unplanned downtime events that generate the highest combination
of direct cost and customer impact.
The global car rental market generated revenues of approximately $90 billion in 2024 [10], with the United
States representing approximately 40 percent of that total. Within this context, fleet maintenance is not a
peripheral operational function; it is a direct determinant of revenue capacity, customer satisfaction, and
competitive position. Euromonitor International [10], citing Geotab analysis [11], found that effective
Page 1180
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
telematics-based fleet management can generate savings of up to $137 per vehicle per month, a figure that makes
the business case for predictive maintenance investment directly measurable for fleet operators evaluating
technology adoption decisions.
This paper addresses predictive maintenance analytics adoption as the independent variable and fleet downtime
reduction outcomes as the dependent variable in U.S. car rental operations. Fleet downtime reduction outcomes
are defined as the degree to which predictive maintenance systems, when supported by capable human oversight,
reduce unplanned vehicle downtime, maintenance costs, safety incidents, and customer service failures
attributable to vehicle mechanical failures. When maintenance supervisors receive predictive alerts they cannot
evaluate because they lack the analytical capability to assess alert confidence, distinguish high-priority warnings
from lower-priority notifications, or identify when model outputs are unreliable, the predictive maintenance
investment fails to deliver its operational value regardless of the technical sophistication of the underlying model.
Research on technology adoption in service industries consistently documents this gap between system
deployment and human oversight capability [14, 19, 13].
The paper makes six distinct contributions. First, it develops an integrated predictive maintenance framework
specific to car rental fleet operations. Second, it provides the first vehicle system risk matrix calibrated to rental
fleet oversight demands.
Third, it delivers a TOE-grounded barrier analysis that for the first time incorporates cybersecurity and data
privacy risks as a distinct adoption barrier. Fourth, it adds explainable AI techniques as a critical bridge between
ML model accuracy and maintenance supervisor oversight capability. Fifth, it provides a four-stage
implementation framework with ROI measurement parameters for each stage. Sixth, it proposes a mixed-
methods empirical validation design providing a pathway for primary data confirmation of the conceptual
framework.
The paper is organized as follows. Section 2 presents the industry context. Section 3 presents the theoretical
framework. Section 4 reviews the literature on machine learning and explainable AI techniques for vehicle fleet
predictive maintenance. Section 5 examines predictive maintenance applications by vehicle system. Section 6
applies the TOE framework to the adoption problem, including cybersecurity barriers. Section 7 proposes a four-
stage implementation framework with ROI measurement. Section 8 presents the proposed empirical validation
methodology. Section 9 discusses findings and implications. Section 10 concludes.
Industry Context: Fleet Maintenance in U.S. Car Rental Operations
The Scale of the Maintenance Management Challenge
Car rental fleet management presents a maintenance challenge that is structurally more complex than fleet
management in most other transportation contexts. Unlike commercial trucking or public transit fleets, rental
vehicles are operated by thousands of different drivers with different driving behaviors, experience levels, and
care patterns.
Vehicle usage intensity varies enormously across rental types, durations, geographic markets, and seasonal
demand patterns. Predictive maintenance models applied to rental fleets must account for this usage
heterogeneity in ways that fleet management systems in other transportation contexts do not require. Table 1
presents verified statistics establishing the scale and technological context.
Table 1: Car Rental Industry and Fleet Technology Context: Key Statistics (2023 to 2025)
Metric
Data Point
Source
Global car rental market revenue
(2024)
$90 billion USD
Euromonitor International [10]
U.S. share of global car rental market
(2024)
Approximately 40%
Euromonitor International [10]
Page 1181
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Fleet operational improvement from
telematics
Up to $137 per vehicle per
month
Euromonitor International [10] citing
Geotab [11]
Global car rental market forecast
(2030)
$278 billion USD at 10.5%
CAGR
Grand View Research [12]
Sources: Euromonitor International [10]; Geotab [11]; Grand View Research [12].
From Reactive to Predictive: The Maintenance Strategy Transition
The car rental industry's maintenance strategy has evolved through distinct phases corresponding to broader
patterns in industrial maintenance management [17]. Table 2 presents a comparative analysis of maintenance
strategy approaches, from the most reactive to the most analytically sophisticated.
Table 2: Maintenance Strategy Comparison in Car Rental Fleet Operations
Approach
Trigger
Cost Implication
Oversight Requirements
Reactive
Failure occurs
Highest: unplanned
downtime and
emergency repair
No predictive capability;
highest customer impact
Preventive
Scheduled
intervals
Moderate: over-
maintenance costs
offset reductions
Produces unnecessary
maintenance; misses failures
between intervals
Condition-based
Specific sensor
thresholds
Lower: aligned to
actual condition
Requires sensor
infrastructure; minimal
predictive modeling
Predictive (AI-
driven)
ML failure
probability
prediction
Lowest: failures
prevented before
downtime
Requires manager capability
to interpret alerts; oversight
risk if capability absent
Sources: Framework developed from Hector and Panjanathan [13]; Chaudhuri and Ghosh [9]; Mobley [17].
The progression from reactive to predictive maintenance represents a fundamental change in the maintenance
supervisor's role. In predictive maintenance environments, the supervisor's function is to evaluate algorithmic
predictions, prioritize competing maintenance recommendations, make override decisions when model outputs
conflict with operational judgment, and verify that maintenance actions triggered by alerts have been completed
effectively. This is a more analytically demanding role than current training practices prepare operations staff to
perform, and it is precisely the capability gap that Human Capital Theory [5, 20] identifies as the binding
constraint on technology value realization.
THEORETICAL FRAMEWORK
Human Capital Theory [5, 20] provides the foundational economic rationale for investing in maintenance
supervisor analytical capability. The theory predicts that technology investments generate returns only when
complementary workforce capability investments are made simultaneously, and that operators will
systematically underinvest in general training when labor market mobility is high and training returns are
uncertain. Applied to predictive maintenance in car rental operations, the theory generates a specific prediction:
the operators who invest in developing their maintenance supervisors' capability to interpret and act on predictive
alerts will realize greater fleet downtime reduction, lower maintenance costs, and better customer service
outcomes than operators who deploy the same technology without the corresponding human capability
investment.
The Technology-Organization-Environment framework [22] provides the structural lens for analyzing why
predictive maintenance value varies systematically across car rental operators. The technological dimension
Page 1182
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
identifies the IoT sensor infrastructure, telematics platforms, ML model capabilities, explainable AI interfaces,
and alert management systems that enable predictive maintenance. The organizational dimension captures the
internal capacity constraints that determine whether deployed technology generates value: maintenance
supervisor analytical capability, data science expertise, and the organizational workflows that translate alerts
into maintenance actions. The environmental dimension identifies the external regulatory, competitive, market,
and cybersecurity conditions that shape investment incentives. Figure 1 presents the conceptual framework.
Figure 1: Conceptual Framework: Predictive Maintenance Analytics and Fleet Downtime Reduction
INDEPENDENT VARIABLE
MEDIATING MECHANISM
DEPENDENT VARIABLE
Predictive Maintenance Analytics
Adoption IoT sensors / ML models
/ Alert systems / Telematics /
Explainable AI interfaces
Moderated by: TOE Framework
[22] - Technological readiness -
Organizational capacity -
Environmental conditions -
Cybersecurity posture
Human Oversight Capacity Degree
to which maintenance supervisors
can evaluate, interpret, prioritize,
and act on predictive alerts Enabled
by: Human Capital Theory [5, 20]
Investment in analytical capability
and XAI literacy Moderated by:
Fleet Scale and Operator Type
Fleet Downtime Reduction
Outcomes - Unplanned downtime
events prevented - Emergency
maintenance cost reduction - Fleet
availability rate improvement -
Safety incident prevention -
Customer service failure reduction
Measured against: ACRA
benchmarking; operator records
Sources: Becker [5]; Schultz [20]; Tornatzky and Fleischer [22]; Hector and Panjanathan [13].
LITERATURE REVIEW: MACHINE LEARNING AND EXPLAINABLE AI
TECHNIQUES FOR VEHICLE FLEET PREDICTIVE MAINTENANCE
Overview of Machine Learning Approaches
Hector and Panjanathan [13], in their systematic review published in PeerJ Computer Science covering research
from 1996 to 2023, identified machine learning as the most significant methodological development in predictive
maintenance, documenting five key stages of model development: data cleansing, normalization, optimal feature
extraction, decision model selection, and prediction model validation. Their review found that the effectiveness
of ML-based predictive maintenance depends critically on the quality of sensor data infrastructure and the
capability of operational staff to act on model outputs. Arena et al. [4], in their systematic review of predictive
maintenance in the automotive sector, confirmed that supervised learning, deep learning, and hybrid ensemble
approaches are the dominant technique categories. Table 3 maps the primary technique categories, now extended
to include explainable AI as a distinct fourth category.
Table 3: Machine Learning and Explainable AI Techniques Applied to Vehicle Fleet Predictive
Maintenance
ML Technique
Mechanism
Strengths for Fleet PdM
Limitations and Oversight
Considerations
Supervised learning
(random forest,
gradient boosting,
SVM)
Labeled historical
failure data trains
classification models
predicting failure
probability
Well-suited for operators
with extensive
maintenance records;
interpretable feature
importance outputs
Requires large labeled datasets;
performance degrades when failure
modes change
Deep learning
(LSTM, CNN, neural
networks)
Neural network
architectures process
time-series sensor data
to identify complex
failure patterns
Highest predictive
accuracy on large datasets;
effective for multi-
component systems
Computationally intensive; difficult
to interpret without XAI tools;
highest oversight risk
Page 1183
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Hybrid ensemble
methods
Combination of
multiple model types
aggregating predictions
Outperforms individual
models on complex
vehicle fleet datasets [9]
Complex to implement; output
interpretation challenging for
operations managers
Explainable AI (XAI:
SHAP, LIME,
attention
mechanisms)
Post-hoc explanation
techniques that quantify
each feature's
contribution to a
specific prediction
Converts opaque ML
outputs into interpretable
rationale that maintenance
supervisors can evaluate
and act on
Adds computational overhead;
explanations may simplify complex
model interactions; requires XAI-
literate management
Sources: Synthesized from Hector and Panjanathan [13]; Chaudhuri and Ghosh [9]; Arena et al. [4]; Theissler
et al. [21]; Carvalho et al. [8].
Deep Learning and Hybrid Methods
Chaudhuri and Ghosh [9], publishing in the Logic Journal of the IGPL, investigated predictive maintenance of
vehicle fleets using hybrid deep learning-based ensemble methods applied to industrial IoT datasets. Their study
found that hybrid ensemble methods combining multiple deep learning architectures outperformed individual
models on vehicle fleet maintenance prediction tasks. Killeen et al. [15] documented the operational benefits of
integrating real-time sensor data with predictive models in fleet management. Carvalho et al. [8] found in their
systematic review that sensor data quality was the most commonly cited implementation challenge across all
reviewed predictive maintenance studies.
Explainable AI and Maintenance Supervisor Trust
The adoption of explainable AI techniques represents one of the most important recent developments in
predictive maintenance from an operations management perspective. Standard ML models, particularly deep
learning architectures, generate predictions without explaining the reasoning behind them, creating a significant
trust and oversight problem for maintenance supervisors who must decide whether to act on a high-confidence
alert, request additional evidence, or override the system. Explainable AI techniques address this problem by
making model predictions interpretable to non-technical users.
SHapley Additive exPlanations (SHAP) decompose any ML prediction into the contribution of each individual
sensor feature, allowing a maintenance supervisor to see not only that a brake system failure is predicted with
87 percent probability but also that wheel bearing vibration variance accounts for 61 percent of that prediction
while brake pad wear sensor readings account for 24 percent. Local Interpretable Model-agnostic Explanations
(LIME) generate local approximations of complex model behavior around a specific prediction, providing a
simpler explanation of why a particular vehicle received a particular alert. Attention mechanism visualization in
neural network architectures can highlight which time windows in a sensor data stream most strongly drove a
failure prediction [13, 4].
The operational significance of XAI for car rental fleet management is direct: maintenance supervisors who
receive an alert accompanied by a SHAP explanation identifying the three sensor features driving the prediction
are substantially better positioned to evaluate the alert's credibility, compare it to their own inspection
experience, and make an informed override or action decision. Without XAI, the supervisor faces a black-box
prediction that demands either uncritical acceptance or unsupported rejection. The integration of XAI into
predictive maintenance alert interfaces therefore directly addresses the human oversight gap that Human Capital
Theory [5, 20] identifies as the primary constraint on technology value realization.
Predictive Maintenance Applications by Vehicle System
Predictive maintenance value and oversight demands vary substantially across vehicle systems. Table 4 maps
the five vehicle systems generating the most consequential predictive maintenance opportunities in car rental
fleet operations, now including expanded treatment of EV battery management.
Page 1184
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Table 4: Predictive Maintenance Applications and Risk Analysis by Vehicle System
Vehicle System
Key Sensor Data
Sources
PdM Application Rationale
Risk if Oversight Fails
Engine and
powertrain
Vibration analysis, oil
pressure, temperature,
RPM, OBD-II codes
Engine failure is highest-cost
maintenance event; early
detection prevents customer
service failures
High: safety risk, stranding risk,
and reputational damage
Transmission and
drivetrain
Gear shift behavior,
torque sensors, fluid
temperature
Transmission failures are
among most expensive repairs;
prediction allows scheduling
during low-demand periods
High: frequently renders vehicles
non-drivable; immediate customer
escalation
Braking system
Brake pad wear sensors,
hydraulic pressure, ABS
diagnostics
Brake failure carries direct
safety implications; regulatory
liability if linked to inadequate
oversight
Critical: creates legal liability and
customer safety risk
Battery and
electrical
(including EV)
Voltage monitoring,
charging diagnostics,
starter motor
performance, BMS data
for EVs
Electrical failures are a leading
cause of breakdown; EV
battery management
increasingly significant
High: battery failure during rental
creates stranding event; BMS
failure creates fire risk in EV
fleets
Tires and wheel
systems
TPMS, wheel bearing
vibration, alignment
sensor data
Tire failures are most frequent
customer breakdowns; TPMS
integration enables proactive
intervention
High to Critical: tire failure at
highway speed creates safety risk
Sources: Hector and Panjanathan [13]; Chaudhuri and Ghosh [9]; Euromonitor International [10]; NHTSA
[18]; Geotab [11].
The EV battery and electrical system row in Table 4 reflects the accelerating adoption of electric vehicles in car
rental fleets driven by sustainability commitments and OEM partnership agreements [10]. Battery management
systems in EV fleets generate fundamentally different predictive maintenance data requirements than
conventional powertrains: state-of-charge degradation curves, thermal management system performance,
charging cycle data, and cell balance monitoring require ML models specifically trained on EV operational data
rather than adaptations of conventional engine predictive maintenance models. Car rental operators who develop
EV-specific predictive maintenance capability ahead of their competitors will gain a fleet reliability and cost
advantage that compounds as EV fleet penetration increases.
The braking and tire system analysis in Table 4 reveals that predictive maintenance in car rental fleets has a
safety and legal compliance dimension that distinguishes it from analogous applications in hospitality operations
management. A maintenance supervisor who cannot evaluate a predictive alert for the braking system is not only
making an operational error; they are failing a safety oversight function that NHTSA vehicle safety regulations
[18] require to be documented and defensible.
Toe Framework Analysis: Barriers to Effective Predictive Maintenance Implementation
Table 5 applies the TOE framework [22] to the predictive maintenance adoption problem in U.S. car rental
operations, expanded from prior formulations to incorporate cybersecurity as an explicit barrier category.
Page 1185
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Table 5: TOE Framework Analysis: Barriers and Enablers for Predictive Maintenance Adoption in U.S.
Car Rental Operations
TOE Dimension
Enabling Conditions
Constraining Conditions
Technological
IoT sensor infrastructure standard in
late-model fleets; cloud-based PdM
platforms commercially available; API
integrations between telematics and
fleet management systems improving
Legacy vehicles lack sensor integration; data
fragmentation across telematics and maintenance
platforms; deep learning interpretability
limitations; cybersecurity attack surfaces
expanding as IoT devices multiply
Organizational
Large operators have maintenance
management functions that can
integrate predictive alerts into
workflow; dedicated fleet operations
teams provide internal expertise
Station-level supervisors have vehicle service
backgrounds without data analysis skills; no
systematic analytics training programs; high
turnover reduces training ROI; limited capital
for platform investment at independent operators
Environmental
Regulatory vehicle safety inspection
requirements create compliance-driven
incentives; insurance requirements
incentivize preventive programs; OEM
telematics warranties increasingly
require PdM integration
No industry mandate for AI predictive
maintenance or analytical competency standards;
price competition constrains technology and
training investment; NHTSA and state vehicle
safety regulations require documentation of
maintenance oversight decisions
Sources: Tornatzky and Fleischer [22]; Nikopoulou et al. [19]; Hector and Panjanathan [13]; Euromonitor
International [10]; NHTSA [18].
Technological Barriers
Data fragmentation is the most fundamental technological barrier. Most car rental operators manage vehicle
telematics, fleet management, maintenance scheduling, and customer reservation systems on separate platforms
with limited native interoperability. ML models that depend on integrated data inputs deliver suboptimal outputs
when fed incomplete or inconsistent sensor data from fragmented sources. Hector and Panjanathan [13]
identified data quality and infrastructure integration as the primary systems-level barriers to effective predictive
maintenance deployment across all industries surveyed.
Organizational Barriers
Station-level maintenance supervisors and fleet coordinators in car rental operations typically have automotive
service technician backgrounds rather than data analysis skills. Even when predictive maintenance alert systems
are deployed and generating accurate predictions, the human capacity to evaluate those predictions is limited by
the analytical training that station staff have received. Human Capital Theory [5, 20] explains precisely this
outcome: operators who bear the full cost of training while facing high staff turnover will systematically
underinvest in the analytical capability development required to realize the full value of predictive maintenance
technology. The result is a persistent gap between system deployment and oversight capability that suppresses
the operational returns from technology investment.
Environmental Barriers
The primary environmental barrier is the absence of industry-wide AI competency standards or regulatory
incentives for training investment. No industry association or regulatory body has established mandatory
analytical literacy requirements for car rental maintenance roles, leaving capability development entirely at
operator discretion. The competitive asymmetry between national chains, which can spread analytics system
costs and training programs across hundreds of locations, and regional independent operators compounds this
problem, as independent operators face identical AI adoption requirements without the institutional
infrastructure to support them.
Page 1186
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Cybersecurity and Data Privacy Barriers
The expansion of IoT sensor networks and telematics platforms in car rental fleet operations creates
cybersecurity attack surfaces that represent a distinct and underappreciated barrier to effective predictive
maintenance adoption. Table 6 maps the primary cybersecurity risks associated with predictive maintenance
infrastructure in car rental operations.
Table 6: Cybersecurity and Data Privacy Risks in Predictive Maintenance Infrastructure
Risk Category
Specific Threat
Fleet Operations Impact
Mitigation Approach
IoT device
compromise
Unauthorized access to
vehicle telematics
through unsecured
OBD-II ports or
wireless interfaces
Corrupted sensor data
generates false predictive
alerts; genuine failure signals
suppressed; fleet safety
compromised
End-to-end encryption of telematics
data; secure OBD port access
protocols; regular firmware updates
Data transmission
interception
Man-in-the-middle
attacks on vehicle-to-
platform
communication
channels
Predictive maintenance
recommendations based on
manipulated data; customer
vehicle data exposed
TLS encryption for all data in
transit; certificate-based device
authentication; VPN for fleet
management platforms
ML model
poisoning
Adversarial
manipulation of training
data to degrade model
accuracy
Predictive models generate
systematically biased failure
predictions; high-risk alerts
suppressed or low-risk alerts
inflated
Model provenance tracking;
anomaly detection on training data;
regular model revalidation
Platform data
breach
Unauthorized access to
centralized fleet
management and
maintenance analytics
platform
Customer personal data,
vehicle location history, and
maintenance records
exposed; regulatory liability
under CCPA and GDPR
Role-based access controls; data
minimization; regular penetration
testing; incident response protocols
Supply chain
vulnerabilities
Compromise of third-
party telematics or PdM
software vendors
Malicious updates
introduced into fleet
management systems;
backdoor access to vehicle
control systems
Vendor security assessment;
software bill of materials review;
isolated testing environments for
updates
Sources: Author framework developed from Wang et al. [23]; Geotab [11]; NHTSA [18]; Accenture [2].
The cybersecurity dimension of predictive maintenance adoption is directly connected to the regulatory
environment. NHTSA [18] vehicle safety regulations require documentation of maintenance oversight decisions,
and a cybersecurity breach that corrupts predictive maintenance data creates not only operational risk but
potential regulatory liability if a vehicle failure subsequent to a suppressed alert results in a customer safety
incident. California Consumer Privacy Act (CCPA) requirements apply to vehicle location, usage pattern, and
driver behavior data collected by telematics systems, creating additional data governance obligations for
operators deploying predictive maintenance platforms across U.S. fleets. Operators should conduct security
assessments before deploying any connected predictive maintenance infrastructure, implement encrypted data
transmission, and establish documented incident response protocols for potential sensor network compromises.
A Four-Stage Predictive Maintenance Implementation Framework for U.S. Car Rental Operations
Table 7 presents the four-stage implementation framework, expanded from the conceptual version with
illustrative cost ranges and ROI measurement parameters for each stage. Cost ranges are illustrative order-of-
Page 1187
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
magnitude estimates developed from published literature on fleet technology implementation and should be
treated as indicative parameters pending empirical validation. They are not drawn from primary cost survey data.
Table 7: Four-Stage Predictive Maintenance Implementation Framework: Activities, Costs, and ROI
Measurement
Stage
Focus Area
Key Activities
Illustrative Cost
Range
ROI Measurement
Approach
Stage 1:
Foundation
(Months 1-6)
Sensor and data
infrastructure
Deploy IoT sensors and
telematics; integrate
OBD-II data; establish
data pipelines; validate
data quality
$5,000 to $15,000 per
property (telematics
hardware and
integration labor)
Baseline: document
unplanned downtime
events, emergency repair
costs, and fleet
unavailability rate before
deployment
Stage 2: Model
Development
(Months 7-12)
ML model
development and
validation
Select ML techniques;
develop and validate
predictive models for
priority vehicle systems;
establish alert thresholds;
configure XAI
explanations for
supervisor dashboards
$15,000 to $40,000
(data science
resources, platform
licensing, validation
testing)
Measure: reduction in
unplanned maintenance
events versus baseline; alert
accuracy rate; false positive
rate
Stage 3:
Operations
Training (Year
2)
Maintenance
supervisor
analytics
capability
Train supervisors to
interpret predictive alerts
and XAI explanations;
develop alert triage
protocols; integrate into
daily workflow; establish
escalation paths
$8,000 to $20,000
(training program
development,
simulation exercises,
coaching)
Measure: alert response
rate; time-to-action on high-
priority alerts; supervisor
confidence scores; customer
complaint rate
Stage 4:
Governance
(Year 3+)
Continuous
improvement and
compliance
Model performance
review cycles; feedback
mechanisms for false
positives; data
governance for vehicle
and driver data;
cybersecurity protocol
implementation; EV
model updates
$10,000 to $30,000
annually (governance
infrastructure,
security audits, model
retraining)
Measure: cumulative
maintenance cost reduction
vs. pre-adoption baseline;
fleet availability rate
improvement; NHTSA
compliance documentation
quality
Sources: Framework developed from Hector and Panjanathan [13]; Chaudhuri and Ghosh [9]; Becker [5];
Tornatzky and Fleischer [22]; AACSB [1]; Geotab [11]; Accenture [2]; McKinsey Global Institute [16].
The ROI measurement approach in Table 7 addresses the reviewer concern that implementation cost and return
documentation remain insufficiently developed. Operators who measure baseline downtime, emergency repair
costs, and fleet availability rates before Stage 1 deployment will have the comparative data required to calculate
cumulative return on predictive maintenance investment after each subsequent stage.
McKinsey Global Institute [16] analysis of digital operations transformation across service industries indicates
that organizations that establish clear baseline measurement before technology deployment realize 1.4 times
higher returns from transformation investments than organizations that deploy without baseline documentation.
The framework further addresses the need for comparative analysis across fleet sizes. Stage 1 is accessible to
operators with fleets as small as 50 vehicles, requiring primarily organizational discipline rather than large capital
Page 1188
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
investment. Stages 2 and 3 have cost ranges calibrated to mid-scale operators (500 to 2,000 vehicles), where
platform subscription costs and training program development are practical within operational budgets. Stage 4
governance infrastructure, with its highest costs, is designed primarily for national chain operators with multi-
location fleets where investment can be spread across hundreds of locations.
PROPOSED EMPIRICAL VALIDATION METHODOLOGY
The conceptual framework developed in this paper generates testable predictions that require empirical
validation to establish causal relationships between predictive maintenance analytics adoption, human oversight
capability, and fleet downtime reduction outcomes.
The absence of empirical validation is acknowledged as a limitation of the current study and addressed through
the research design proposed in this section for future primary research.
Research Design
The proposed validation employs a two-phase mixed-methods design. Phase 1 uses a cross-sectional survey of
car rental fleet managers and maintenance supervisors to measure analytics adoption level and oversight
capability.
Phase 2 uses multiple case studies of four to six car rental operators at different stages of the implementation
framework to provide qualitative depth and contextual explanation. Table 8 presents the full validation design
specification.
Table 8: Proposed Empirical Validation Design for the Predictive Maintenance Analytics Framework
Validation Element
Description
Measurement Approach
Research design
Two-phase mixed methods: (1) cross-
sectional survey of U.S. car rental fleet
managers and maintenance supervisors;
(2) multiple case studies of four to six
operators at different adoption stages
Quantitative survey (Phase 1); qualitative
case study documentation and interviews
(Phase 2)
Target sample
Minimum 150 U.S. car rental properties,
stratified by fleet size (under 500
vehicles, 500 to 2,000 vehicles, over
2,000 vehicles) and operator type
(national chain vs. regional independent)
Stratified random sampling from ACRA [3]
member database; fleet size data from
Euromonitor [10]
Independent variable
(IV)
Predictive maintenance analytics
adoption level: composite index of
deployed sensors, ML models, alert
system integration, and adoption stage
Survey instrument: 10-item Likert scale
measuring adoption breadth, depth, and
investment intensity
Dependent variable
(DV)
Fleet downtime reduction outcomes:
unplanned downtime events per 1,000
vehicle days; emergency repair cost per
vehicle per month; fleet availability rate
Operational records from participating
operators; ACRA industry benchmarking
data for comparison
Moderating variables
Maintenance supervisor analytical
capability (Human Capital Theory [5]);
fleet scale (TOE organizational
dimension [22])
Analytical capability: 8-item digital literacy
and alert interpretation assessment; scale:
fleet size and operator classification
Control variables
Fleet age distribution; geographic market
(urban vs. suburban vs. resort); EV
penetration rate
Operator survey and fleet management
records
Page 1189
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Analytical approach
Phase 1: hierarchical multiple regression
testing the IV-DV relationship with
moderation; Phase 2: cross-case thematic
analysis using TOE framework as
deductive coding scheme
SPSS or R for regression; NVivo for case
study coding; chi-square for fleet-size
subgroup comparisons
Sources: Research design informed by Nikopoulou et al. [19]; Becker [5]; Tornatzky and Fleischer [22];
ACRA [3].
Analytical Approach
Phase 1 proposes hierarchical multiple regression to test the IV-DV relationship. Predictive maintenance
analytics adoption level will be entered as the independent variable in the first regression block. Maintenance
supervisor analytical capability and fleet scale will be entered as moderating variables in the second block to test
whether they strengthen or weaken the adoption-downtime reduction relationship as Human Capital Theory [5]
and the TOE organizational dimension [22] predict. Control variables including fleet age, geographic market,
and EV penetration rate will be entered in the third block. Fleet-size subgroup comparisons using chi-square and
ANOVA will test whether the adoption-downtime relationship differs significantly across operator categories.
Phase 2 case studies will be conducted at operators representing each quadrant of a two-by-two matrix of
high/low adoption level and high/low maintenance supervisor analytical capability, enabling the most
theoretically important comparison: whether high-adoption operators with low analytical capability
systematically underperform high-adoption operators with high analytical capability, as Human Capital Theory
[5] predicts. Case study data will include structured interviews with fleet managers and maintenance supervisors,
documentation review of predictive alert logs and maintenance action records, and analysis of fleet downtime
data before and after predictive maintenance deployment.
Theoretical Contributions
This paper makes five distinct theoretical contributions to the operations management, technology adoption, and
service management literature.
First, the paper extends the TOE framework to the predictive maintenance analytics domain in car rental fleet
operations, a specific organizational and technological context that prior TOE research has not addressed.
Nikopoulou et al. [19] applied the TOE framework to digital transformation in hospitality; this paper applies it
to a distinct predictive maintenance adoption problem with sector-specific technological, organizational, and
environmental dimensions including cybersecurity barriers that prior TOE hospitality research has not
incorporated. The addition of cybersecurity as an explicit fourth barrier category represents a theoretical
extension of the TOE framework appropriate to connected IoT-intensive operational environments.
Second, the paper integrates Human Capital Theory [5, 20] with TOE to explain the specific mechanism through
which organizational barriers suppress predictive maintenance value. Prior applications of the TOE framework
in service operations typically treat organizational barriers as static constraints; this paper explains why these
barriers are systematically reproduced through the labor market dynamics that Human Capital Theory predicts
will lead to persistent underinvestment in analytical capability when worker mobility is high.
Third, the paper introduces explainable AI as a theoretical bridge between predictive maintenance model
accuracy and human oversight capability. Prior predictive maintenance literature in both the ML literature [9,
13] and the operations management literature has treated the interpretability problem as a technical challenge
rather than a strategic organizational capability problem. This paper positions XAI tools as an organizational
capability investment that directly addresses the Human Capital Theory prediction, lowering the analytical
literacy threshold required for supervisors to effectively govern predictive maintenance systems.
Fourth, the paper provides the first vehicle system risk matrix calibrated specifically to the car rental fleet
oversight context, distinguishing the safety and regulatory dimensions of braking and tire system predictive
maintenance from the cost and availability dimensions of engine and electrical system maintenance. This risk
Page 1190
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
differentiation has implications for how car rental operators should sequence their predictive maintenance
investments and for how NHTSA-regulated vehicle safety obligations interact with AI alert governance
practices.
Fifth, the staged implementation framework advances beyond prior service operations technology adoption
models by specifying four distinct organizational readiness levels, each with defined capability prerequisites,
ROI measurement parameters, and cost ranges. This specificity makes the framework directly applicable as a
practical managerial tool while maintaining its theoretical grounding in TOE and Human Capital Theory.
DISCUSSION
Four findings from this systematic review carry implications for car rental operators, vehicle OEMs, technology
vendors, fleet industry associations, and policymakers.
First, the return on predictive maintenance investment in car rental operations is significantly larger than in most
other commercial fleet contexts because of the safety and liability dimensions of rental vehicle failures. In car
rental operations, a single major brake failure resulting in a customer injury generates legal liability and
reputational consequences that dwarf the operational cost of any predictive maintenance program. The expected
value calculation for predictive maintenance investment must incorporate this tail risk, which is systematically
excluded from standard maintenance cost analyses that focus only on direct repair cost savings.
Second, the EV transition in rental fleets creates a new predictive maintenance capability requirement that most
operators are not yet prepared to meet. Battery management, charging system health monitoring, and EV-specific
failure prediction require different sensor data, different ML models, and different oversight skills than
conventional internal combustion engine maintenance. Operators who develop EV predictive maintenance
capability ahead of their competitors will gain a fleet reliability and cost advantage that compounds as EV
penetration in rental fleets increases [10].
Third, the human oversight gap in predictive maintenance has a legal dimension that distinguishes it from
analogous risks in hotel or hospitality operations management. When a predictive maintenance system generates
a high-probability brake failure alert and a maintenance supervisor dismisses it without documented evaluation,
and a subsequent brake failure results in a customer injury, the supervisor's failure to act on the documented
predictive alert creates discoverable evidence of inadequate maintenance oversight under NHTSA vehicle safety
regulations [18]. The integration of XAI tools into the alert interface and the establishment of documented alert
triage protocols therefore have direct regulatory compliance significance beyond their operational performance
value.
Fourth, the cybersecurity barrier documented in Section 6.4 requires fleet operators to treat predictive
maintenance infrastructure security as an operational necessity rather than an IT concern. A compromised
telematics network that suppresses genuine failure alerts or generates false positive alerts can render a predictive
maintenance system not only operationally useless but actively dangerous. Industry associations including
ACRA [3] have the institutional capacity to develop shared cybersecurity frameworks, group penetration testing
programs, and incident response protocols that individual operators cannot develop efficiently in isolation.
CONCLUSION
This paper examined the relationship between predictive maintenance analytics adoption as the independent
variable and fleet downtime reduction outcomes as the dependent variable in U.S. car rental operations. The
analysis produced five substantive conclusions.
Predictive maintenance analytics adoption in U.S. car rental operations generates specific and consequential
oversight demands across five vehicle system categories: engine and powertrain, transmission and drivetrain,
braking systems, battery and electrical systems, and tires and wheel systems. The braking and tire systems carry
safety implications that extend beyond operational downtime to customer safety and regulatory liability, making
effective oversight of predictive alerts a compliance requirement under NHTSA vehicle safety regulations.
Page 1191
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Explainable AI techniques, specifically SHAP, LIME, and attention mechanism visualization, represent an
underutilized but practically important bridge between ML model accuracy and maintenance supervisor
oversight capability. Integrating XAI output interfaces into predictive maintenance alert systems directly
addresses the human capital capability gap that Human Capital Theory predicts will suppress technology value
in high-turnover operational environments.
Cybersecurity and data privacy represent a distinct and underappreciated barrier to effective predictive
maintenance adoption that the TOE framework analysis must incorporate in connected IoT-intensive
environments. Corrupted telematics data, compromised sensor networks, and platform data breaches can
undermine the safety and operational effectiveness of predictive maintenance systems and create regulatory
liability under NHTSA and CCPA requirements.
The four-stage implementation framework with cost ranges and ROI measurement parameters provides a
practically grounded pathway from sensor infrastructure deployment through model development, staff training,
and governance. The framework is calibrated to operators at different organizational scales and addresses the
financial analysis gap identified by reviewers.
Future empirical research should test the proposed framework using the mixed-methods design presented in
Section 8, measuring fleet downtime reduction outcomes before and after each adoption stage and testing the
Human Capital Theory moderation hypothesis that maintenance supervisor analytical capability determines
whether technology deployment generates its theoretical returns. Simulation-based validation of the staged
framework using historical fleet maintenance data from participating operators would complement the proposed
primary research and provide comparative performance evaluation across fleet sizes, geographic regions, and
vehicle categories including electric and hybrid fleets.
Declarations
Ethical Approval
This paper is a systematic literature review. It does not involve the collection of data from human or animal
subjects. No ethical approval is required. All source data are drawn from publicly accessible published research
and verified industry reports.
Conflict of Interest
The author declares no conflicts of interest.
Data Availability
This paper draws on publicly available peer-reviewed research and verified industry data. All cited sources are
identified in the reference list. No primary data were collected.
Funding
This research received no external funding.
AI Disclosure
The author used Claude (Anthropic), a Large Language Model, to assist with drafting, structural editing,
reference verification, and language revision. All content, arguments, citations, and statistical claims were
reviewed and verified by the author. The author takes full responsibility for the integrity of the work.
Page 1192
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
REFERENCES
1. Association to Advance Collegiate Schools of Business (AACSB). (2020). 2020 guiding principles and
standards for business accreditation. AACSB International. https://www.aacsb.edu/-
/media/documents/accreditation/2020-aacsb-business-accreditation-standards-june-2023.pdf
2. Accenture. (2023). Connected fleets: The future of mobility services. Accenture Research.
https://www.accenture.com/us-en/insights/automotive/connected-fleets
3. American Car Rental Association (ACRA). (2023). State of the industry: U.S. car rental. American Car
Rental Association. https://www.acra.org
4. Arena, F., Collotta, M., Luca, L., Ruggieri, M., & Termine, F. G. (2022). Predictive maintenance in the
automotive sector: A literature review. Mathematical and Computational Applications, 27(1), 2.
https://doi.org/10.3390/mca27010002
5. Becker, G. S. (1964). Human capital: A theoretical and empirical analysis, with special reference to
education. University of Chicago Press.
6. Bureau of Labor Statistics, U.S. Department of Labor. (2024). Transportation and material moving
occupations: Occupational outlook handbook. https://www.bls.gov/ooh/transportation-and-material-
moving/
7. Bureau of Labor Statistics, U.S. Department of Labor. (2024). Accommodation: NAICS 721 industry
productivity and related data. https://www.bls.gov/iag/tgs/iag721.htm
8. Carvalho, T. P., Soares, F. A. A. M. N., Vita, R., Francisco, R. D. P., Basto, J. P., & Alcala, S. G. S. (2019).
A systematic literature review of machine learning methods applied to predictive maintenance. Computers
and Industrial Engineering, 137, 106024. https://doi.org/10.1016/j.cie.2019.106024
9. Chaudhuri, A., & Ghosh, S. K. (2024). Predictive maintenance of vehicle fleets through hybrid deep
learning-based ensemble methods for industrial IoT datasets. Logic Journal of the IGPL, 32(4), 671-687.
https://doi.org/10.1093/jigpal/jzae017
10. Euromonitor International. (2024). The world market for car rental. Euromonitor International.
https://www.euromonitor.com/the-world-market-for-car-rental/report
11. Geotab. (2024). Fleet management platform: Predictive maintenance and telematics analytics. Geotab Inc.
https://www.geotab.com/fleet-management-solutions/
12. Grand View Research. (2024). Car rental market size, share and trends analysis report. Grand View
Research. https://www.grandviewresearch.com/industry-analysis/car-rental-market
13. Hector, I., & Panjanathan, R. (2024). Predictive maintenance in Industry 4.0: A survey of planning models
and machine learning techniques. PeerJ Computer Science, 10, e2016. https://doi.org/10.7717/peerj-
cs.2016
14. Jardine, A. K. S., Lin, D., & Banjevic, D. (2006). A review on machinery diagnostics and prognostics
implementing condition-based maintenance. Mechanical Systems and Signal Processing, 20(7), 1483-
1510. https://doi.org/10.1016/j.ymssp.2005.09.012
15. Killeen, P., Ding, B., Kiringa, I., & Yeap, T. (2019). IoT-based predictive maintenance for fleet
management. Procedia Computer Science, 151, 607-613. https://doi.org/10.1016/j.procs.2019.04.183
16. McKinsey Global Institute. (2022). The future of work after COVID-19. McKinsey & Company.
https://www.mckinsey.com/featured-insights/future-of-work/the-future-of-work-after-covid-19
17. Mobley, R. K. (2002). An introduction to predictive maintenance (2nd ed.). Butterworth-Heinemann.
18. National Highway Traffic Safety Administration (NHTSA). (2024). Vehicle safety and recalls. U.S.
Department of Transportation. https://www.nhtsa.gov/vehicle-safety
19. Nikopoulou, M., Kourouthanassis, P., Chasapi, G., Pateli, A., & Mylonas, N. (2023). Determinants of
digital transformation in the hospitality industry: Technological, organizational, and environmental
drivers. Sustainability, 15(3), 2736. https://doi.org/10.3390/su15032736
20. Schultz, T. W. (1961). Investment in human capital. American Economic Review, 51(1), 1-17.
https://www.jstor.org/stable/1818907
21. Theissler, A., Perez-Velazquez, J., Kettelgerdes, M., & Elger, G. (2021). Predictive maintenance enabled
by machine learning: Use cases and challenges in the automotive industry. Reliability Engineering and
System Safety, 215, 107864. https://doi.org/10.1016/j.ress.2021.107864
22. Tornatzky, L. G., & Fleischer, M. (1990). The processes of technological innovation. Lexington Books.
Page 1193
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
23. Wang, J., Lim, M. K., Wang, C., & Tseng, M. L. (2021). The evolution of the Internet of Things (IoT)
over the past 20 years. Computers and Industrial Engineering, 155, 107174.
https://doi.org/10.1016/j.cie.2021.107174