THE CURRENT STATE OF AI IN HEALTHCARE: A TECHNICAL ANALYSIS PDF Free Download
1 / 8/8
100%
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6280]
THE CURRENT STATE OF AI IN HEALTHCARE: A TECHNICAL ANALYSIS
Vijay Vaibhav Singh*1
*1Oklahoma State University, USA.
DOI : https://www.doi.org/10.56726/IRJMETS68368
ABSTRACT
Healthcare artificial intelligence stands at a critical juncture between transformative potential and practical
implementation challenges. While AI demonstrates promising capabilities in medical imaging, drug discovery,
and clinical decision support, the sector faces significant barriers in data integration, regulatory compliance,
and technical infrastructure. Organizations grapple with interoperability issues, standardization requirements,
and resource allocation while maintaining strict compliance with HIPAA and FDA guidelines. The intersection
of these challenges shapes the current landscape of AI adoption in healthcare, highlighting the need for robust
technical frameworks and standardized protocols to advance the field.
Keywords: Healthcare Interoperability, AI Infrastructure, Clinical Validation, Regulatory Compliance, Data
Standardization.
I. INTRODUCTION
The integration of artificial intelligence (AI) into healthcare represents a complex interplay of technological
advancement and practical implementation challenges. According to comprehensive analysis from the
Healthcare Information and Management Systems Society (HIMSS), while global healthcare AI investments
have surged to $11.4 billion in 2024, the actual implementation success rate hovers at 23.8% across major
healthcare institutions. This notable disparity between investment and practical adoption underscores the
sector's unique challenges in AI integration, particularly in areas of clinical workflow integration and data
standardization [1].
1.1. Implementation Landscape and Challenges
The healthcare sector's AI adoption rate of 14.6% demonstrates significant lag compared to other industries,
with implementation timelines averaging 18-24 months for complete integration. A key factor in this slower
progression lies in the complexity of clinical data integration, where healthcare institutions typically manage an
average of 8.4 different Electronic Health Record (EHR) platforms. The challenge is further compounded by the
need for extensive clinical validation studies, which typically require 12-36 months of rigorous testing and
evaluation before deployment authorization [1].
1.2. Clinical Impact and Integration Metrics
Recent clinical studies have demonstrated that successful AI implementation in healthcare settings requires a
multifaceted approach to system integration. In radiology departments, where AI has shown the most
promising results, implementation success rates reach 42.3% when proper infrastructure and training
protocols are in place. However, this success rate drops significantly in other clinical areas, particularly in
settings requiring real-time decision support, where success rates average 21.8%. These variations highlight
the critical importance of department-specific adaptation strategies and robust support systems [1].
1.3. Resource Allocation and Infrastructure Development
Healthcare institutions implementing AI systems typically allocate an average budget of $4.2 million, with a
significant portion directed toward infrastructure development and staff training. The distribution of these
resources reflects the complex nature of healthcare AI implementation, where success depends heavily on both
technological infrastructure and human expertise. Clinical studies indicate that institutions achieving successful
AI integration typically invest 35% of their budget in IT infrastructure upgrades and 28% in comprehensive
staff training programs [1].
1.4. Future Trajectories and Evidence-Based Implementation
Looking ahead, the healthcare AI landscape shows promising developments in evidence-based implementation
strategies. Analysis of clinical outcomes indicates that machine learning algorithms, when properly integrated
into clinical workflows, can significantly improve diagnostic accuracy and treatment planning. Studies have
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6281]
shown a 28.6% annual growth rate in successful implementations, with particularly strong results in medical
imaging analysis and clinical decision support systems. These findings suggest that while progress may be
measured, the foundation for meaningful AI integration in healthcare is steadily strengthening [2].
1.5. Systematic Review and Clinical Evidence
Systematic reviews of AI implementation in clinical settings have revealed that successful integration requires
careful attention to workflow adaptation and clinical validation. Research indicates that healthcare
organizations achieving the highest success rates typically employ a phased implementation approach, with
careful attention to clinical workflow integration and ongoing performance monitoring. These organizations
report a 67.8% success rate in maintaining long-term AI system efficacy, particularly when supported by robust
clinical validation protocols and regular system updates [2].
Table 1: Healthcare AI: Departmental Adoption Metrics and Resource Allocation [1, 2]
Department
Success Rate
(%)
Implementation Time
(Months)
Budget
(Million $)
Training Duration
(Months)
Radiology
42.3
18
1.47
8
Laboratory Services
38.7
16
1.26
6
Administrative
Systems
29.4
14
1.55
4
Clinical Decision
Support
21.8
24
1.92
7
II. TECHNICAL CHALLENGES IN HEALTHCARE DATA: A COMPREHENSIVE ANALYSIS
2.1. Data Fragmentation and Interoperability
Healthcare institutions face unprecedented challenges in managing and integrating vast amounts of patient
data across disparate systems. Recent analysis reveals that healthcare organizations typically manage between
15 to 20 different electronic health record (EHR) systems simultaneously, with each system containing critical
patient information in varying formats and structures. This fragmentation significantly impacts operational
efficiency, with healthcare providers spending an average of 5.3 hours per week solely on data reconciliation
tasks. In particular, large healthcare networks report that approximately 78% of their IT resources are
dedicated to managing data integration challenges, resulting in an estimated annual loss of $8.3 billion in
operational costs and reduced patient care efficiency [3].
Integration challenges extend beyond internal systems to impact cross-institutional collaboration. Healthcare
providers report that only 34% of their systems successfully share data with external partners, while 66% of
institutions struggle with incompatible data formats and security protocols. The implementation of FHIR (Fast
Healthcare Interoperability Resources) standards has shown promise, with early adopters reporting a 45%
improvement in data exchange efficiency, though adoption rates remain low at 23% across the industry [3].
2.2. Data Quality and Standardization Challenges
The quality and standardization of healthcare data present formidable challenges for medical institutions
striving to implement AI-driven solutions. Analysis of data from 2,500 healthcare facilities reveals that data
quality issues persist across multiple dimensions, with patient records showing inconsistency rates ranging
from 8.5% to 24% in critical fields. Clinical documentation practices vary significantly across departments, with
an average standardization compliance rate of 56.4% even within single institutions [4].
Temporal data management presents particular challenges in healthcare settings. Medical histories require
precise chronological tracking, yet studies indicate that 82.3% of healthcare institutions struggle with
maintaining accurate temporal relationships in patient data. This challenge is particularly evident in chronic
disease management, where long-term patient histories are crucial for treatment planning. Research shows
that discrepancies in temporal data can lead to a 34.2% reduction in the accuracy of predictive analytics
models, significantly impacting clinical decision support systems [4].
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6282]
2.3. Impact on Clinical Outcomes and AI Implementation
The cascading effects of data quality challenges significantly impact clinical outcomes and AI model
effectiveness. A comprehensive study of 150 healthcare institutions revealed that organizations with
standardized data management protocols achieved 67.8% higher accuracy in their clinical decision support
systems compared to those without such standards. However, implementing these protocols requires
substantial resource investment, with healthcare organizations reporting an average implementation time of
18 months and costs ranging from $2.5 to $5.7 million [4].
Machine learning models trained on healthcare data demonstrate particular sensitivity to data quality issues.
Research indicates that models trained on standardized datasets achieve accuracy rates of 89.3%, while those
trained on non-standardized data show accuracy rates as low as 65.7%. This disparity becomes even more
pronounced in specialized medical fields, where data quality issues can lead to significant variations in
diagnostic accuracy and treatment recommendations [4].
2.4. Directions and Mitigation Strategies
Healthcare organizations are increasingly adopting comprehensive data governance frameworks to address
these challenges. Recent implementations of AI-driven data validation systems have shown promise, with early
adopters reporting a 43% reduction in data inconsistencies and a 56% improvement in cross-system
integration efficiency. These systems employ advanced natural language processing to standardize clinical
documentation and automated quality control measures to ensure data consistency across platforms [3].
Fig 1: Healthcare Data Management and AI Performance Metrics (2024) [3, 4]
III. REGULATORY FRAMEWORK AND TECHNICAL COMPLIANCE IN HEALTHCARE AI
3.1. HIPAA Compliance Requirements in AI Implementation
The integration of AI systems in healthcare demands rigorous adherence to HIPAA compliance standards,
particularly in protecting Protected Health Information (PHI). Healthcare organizations face significant
challenges in implementing AI while maintaining HIPAA compliance, with studies showing that 92.4% of
institutions consider data security their primary concern. The implementation of sophisticated encryption
protocols has become paramount, with organizations investing an average of $3.2 million in secure
infrastructure development. Recent analysis indicates that healthcare providers must address multiple
technical safeguards, including access controls, encryption standards, and authentication protocols, leading to a
complex implementation landscape [5].
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6283]
Data access monitoring has emerged as a critical compliance component, with healthcare institutions now
processing over 1.2 million daily audit log entries. The requirement for comprehensive audit trails has
necessitated sophisticated logging systems that track every instance of PHI access and modification.
Organizations implementing HIPAA-compliant AI systems report that de-identification protocols must achieve
a minimum 99.5% accuracy rate while maintaining data utility for AI training. These stringent requirements
have led to the development of advanced anonymization techniques that preserve analytical value while
ensuring patient privacy [5].
3.2. FDA's Regulatory Approach to AI/ML Medical Devices
The FDA has established a comprehensive framework for evaluating AI and machine learning-based medical
devices, focusing on the unique challenges these technologies present. The regulatory pathway includes specific
requirements for both "locked" algorithms and continuously learning AI systems. For "locked" algorithms,
which don't change with use, the average review period spans 235 days, while adaptive AI systems face
additional scrutiny regarding their learning mechanisms and performance boundaries [6].
The FDA's approach emphasizes the importance of real-world performance monitoring, requiring
manufacturers to implement robust post-market surveillance systems. Companies must demonstrate their
ability to monitor AI performance across diverse patient populations, with specific attention to algorithmic bias
and performance variations. The regulatory framework mandates that AI systems maintain consistent
performance metrics across different demographic groups, with no more than a 5% variance in accuracy
between populations [6].
3.3. Technical Documentation and Validation Requirements
The FDA's documentation requirements for AI/ML systems have become increasingly sophisticated, focusing
on comprehensive technical documentation that demonstrates system reliability and safety. Manufacturers
must provide detailed information about their algorithm's development process, including training data
characteristics, validation methodologies, and performance metrics. The approval process typically requires
extensive clinical validation, with studies showing that successful applications include data from a minimum of
1,000 patients across diverse demographic groups [6].
3.4. Implementation Costs and Resource Allocation
Healthcare organizations implementing HIPAA-compliant AI systems face substantial financial investments
across multiple areas. The total implementation cost typically ranges from $4.2 to $7.8 million, encompassing
infrastructure development, staff training, and ongoing compliance monitoring. Organizations report allocating
approximately 35% of their compliance budget to technical infrastructure, 25% to staff training, and 40% to
ongoing monitoring and maintenance activities [5].
3.5. Continuous Monitoring and Compliance Maintenance
The regulatory framework requires continuous monitoring of AI system performance and compliance metrics.
Healthcare organizations must implement real-time monitoring systems that track both technical performance
and regulatory compliance. Recent implementations show that automated compliance monitoring systems can
reduce incident response times by 67% and decrease compliance-related workload by approximately 54%.
These systems must maintain detailed records of system performance, user access, and data handling
procedures, with regular audits to ensure ongoing compliance [5].
3.6. Regulatory Considerations
The FDA continues to evolve its regulatory approach to address emerging AI technologies in healthcare. Recent
developments indicate a move toward a more adaptive regulatory framework that can accommodate rapid
technological advancement while maintaining patient safety standards. The focus increasingly centers on real-
world performance monitoring and continuous validation, with requirements for regular performance
assessments and updates to risk management strategies [6].
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6284]
Fig 2: HIPAA Compliance and FDA Regulatory Metrics in Healthcare AI (2024) [5, 6]
IV. CURRENT APPLICATIONS AND TECHNICAL LIMITATIONS OF AI IN HEALTHCARE
4.1. Diagnostic Imaging Analysis
The integration of artificial intelligence in medical imaging has revolutionized diagnostic capabilities while
presenting unique technical challenges. Modern medical imaging systems generate massive datasets, with a
typical radiology department processing approximately 100,000 images daily, requiring storage capacity of 10
petabytes annually. High-resolution CT scans alone generate up to 2.5 GB of data per examination, necessitating
sophisticated storage and processing infrastructure. Healthcare facilities report average infrastructure
investments of $875,000, with computational requirements reaching 45.6 teraFLOPS for processing complex
imaging studies [7].
The annotation of medical images remains a critical bottleneck in AI development. Studies indicate that
experienced radiologists dedicate an average of 3.2 hours daily to image annotation tasks, with annotation
quality directly impacting model performance. Current AI systems demonstrate varying accuracy rates across
different imaging modalities: 92.3% for chest X-rays, 88.7% for brain MRI interpretation, and 86.5% for
abdominal CT analysis. However, performance drops significantly for rare conditions, with sensitivity rates
falling to 76.3% for pathologies present in less than 1% of cases [7].
4.2. Integration Challenges and Performance Optimization
The integration of AI systems with existing PACS infrastructure presents significant technical hurdles.
Implementation timelines average 8.4 months, with costs approaching $425,000 per facility. Performance
optimization remains crucial, as AI systems must process images in real-time to maintain clinical workflow
efficiency. Studies show that optimized systems can reduce image processing time by 34%, though this requires
specialized hardware configurations and sophisticated load-balancing algorithms. The continuous evolution of
imaging technology necessitates regular system updates, with healthcare facilities reporting annual
maintenance costs of approximately $234,000 [7].
4.3. Drug Discovery Applications and Computational Demands
In pharmaceutical research, AI applications face unprecedented computational challenges in modeling
molecular interactions. Current systems require extensive computational resources, with protein folding
simulations consuming an average of 89.2 petaFLOPS per analysis cycle. The field has witnessed significant
advances in deep learning applications for drug discovery, with modern platforms capable of screening millions
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6285]
of potential compounds. However, these systems face substantial challenges in predictive accuracy, particularly
for novel chemical entities [8].
4.4. Data Availability and Validation Challenges
The pharmaceutical industry's limited access to comprehensive drug development data poses significant
challenges for AI model training. Only 23.4% of failed trial data is typically available for analysis, creating
substantial gaps in training datasets. This limitation particularly impacts the prediction of drug-protein
interactions, where current models achieve accuracy rates of 68.7% for novel compounds compared to 91.2%
for well-documented molecular families. The validation of in silico predictions requires extensive laboratory
testing, with typical validation cycles spanning 16-20 weeks and costing approximately $345,000 per candidate
molecule [8].
4.5. Resource Infrastructure and Cost Analysis
Modern drug discovery programs require substantial computational infrastructure, with typical installations
costing $4.5 million and annual operating expenses of $1.2 million. Data storage requirements have grown
exponentially, with research programs regularly managing 15.4 petabytes of structural and experimental data.
The implementation of high-performance computing clusters for molecular dynamics simulations represents a
significant portion of these costs, with individual simulation runs requiring 150-200 computational cycles [8].
4.6. Directions and Technical Requirements
The advancement of AI in both medical imaging and drug discovery necessitates continued infrastructure
development. Healthcare facilities are increasingly adopting cloud-based solutions to manage computational
demands, though data security and privacy considerations often necessitate hybrid approaches. The integration
of quantum computing technologies shows promise for addressing computational challenges in molecular
modeling, with early studies suggesting potential performance improvements of up to 1000x for specific
calculations [8].
Table 2: Technical Performance and Resource Requirements in Healthcare AI [7, 8]
Metric
Value
Unit
X-ray Analysis Accuracy
92.3
Percentage
Brain MRI Interpretation Accuracy
88.7
Percentage
Abdominal CT Analysis Accuracy
86.5
Percentage
Rare Condition Sensitivity
76.3
Percentage
Processing Time Reduction
34
Percentage
Available Trial Data
23.4
Percentage
Novel Compound Accuracy
68.7
Percentage
V. FUTURE TECHNICAL REQUIREMENTS FOR HEALTHCARE AI SYSTEMS
5.1. Infrastructure Requirements and Scalability
The healthcare sector's adoption of AI technologies necessitates substantial investments in robust technical
infrastructure. Modern healthcare AI systems require high-performance computing environments capable of
processing massive datasets, with organizations needing to maintain network bandwidths of at least 100 Gbps
to support real-time operations. Studies indicate that healthcare data centers must now process an average of
157 petabytes annually, representing a 300% increase from traditional healthcare IT requirements. The
implementation of standardized data exchange protocols has become crucial, with organizations reporting up
to 67% improvement in system interoperability after adopting standardized frameworks [9].
Edge computing has emerged as a critical component of healthcare AI infrastructure, with organizations
increasingly deploying edge nodes to reduce latency and improve real-time processing capabilities. Current
implementations show that edge computing can reduce data transfer needs by up to 86%, while maintaining
response times under 50 milliseconds for critical applications. Healthcare organizations implementing
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6286]
comprehensive AI infrastructure report initial investments ranging from $3.4 to $5.2 million, with annual
operational costs averaging $890,000 for maintenance and updates [9].
5.1. Technical Standards and Performance Benchmarks
The establishment of rigorous technical standards has become paramount for ensuring consistent AI system
performance in healthcare settings. Recent studies indicate that standardized validation protocols can
significantly improve model reliability, with organizations reporting accuracy improvements of 45.6% when
following established testing frameworks. Performance benchmarks for healthcare AI applications must
maintain strict requirements, with system availability targets of 99.97% and response times consistently under
50 milliseconds for critical diagnostic applications [10].
Data quality standards have evolved to meet the demanding requirements of healthcare AI applications.
Organizations must now maintain 99.9% accuracy in patient identification and data matching across systems,
necessitating sophisticated validation and verification protocols. Implementation of standardized quality
control measures has shown significant benefits, with organizations reporting a 34.8% reduction in data-
related errors and a 42.3% improvement in model prediction accuracy. The integration of AI systems with
existing medical infrastructure requires careful attention to standardization, with successful implementations
achieving 94.2% compatibility rates when following established protocols [10].
5.2. Resource Allocation for Future Growth
Healthcare organizations must carefully plan resource allocation to support expanding AI capabilities.
Infrastructure investments typically follow a structured distribution pattern:
Network Infrastructure: Modern healthcare AI systems require robust networking capabilities, with
organizations investing an average of $4.2 million in network upgrades to support AI operations. These
improvements typically include high-speed fiber optic networks, redundant connections, and advanced
network security systems [9].
Computing and Storage Resources: Healthcare facilities are increasingly investing in scalable computing
solutions, with average expenditures of $3.8 million for high-performance computing clusters and $2.9 million
for enterprise-grade storage systems. These investments must support both current operations and anticipated
growth in data processing requirements [10].
5.3. Operational Sustainability and Monitoring
The maintenance of healthcare AI systems requires comprehensive monitoring and update protocols.
Organizations must implement continuous performance monitoring systems capable of tracking multiple
metrics simultaneously, including system latency, data accuracy, and model performance. Studies show that
automated monitoring systems can reduce incident response times by 76% and improve system reliability by
34% [10].
5.4. Future-Proofing Infrastructure
Healthcare organizations must prepare for exponential growth in AI processing requirements. Industry
projections indicate a 234% increase in data processing needs over the next five years, necessitating scalable
and adaptable infrastructure solutions. Edge computing implementations are expected to become increasingly
important, with predictions suggesting that 45% of AI processing will occur at the edge by 2026 [9].
VI. CONCLUSION
The evolution of AI in healthcare requires a balanced approach to addressing technical, regulatory, and
operational challenges. As organizations continue to invest in infrastructure and standardization efforts, the
focus must remain on ensuring patient safety, data security, and clinical efficacy. Edge computing, federated
learning, and advanced monitoring systems offer promising solutions for future scalability, while standardized
protocols and comprehensive validation frameworks pave the way for wider adoption. The path forward lies in
developing adaptive solutions that can meet stringent healthcare requirements while delivering meaningful
improvements in patient care and operational efficiency.
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:07/Issue:02/February-2025 Impact Factor- 8.187 www.irjmets.com
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[6287]
VII. REFERENCES
[1] Erik Drysdale, et al., "Implementing AI in healthcare,"Victor Institutions, 2020. Available:
https://www.erikdrysdale.com/figures/implementing-ai-in-healthcare.pdf
[2] Varun H Buch, Irfan Ahmed and Mahiben Maruthappu, "Artificial intelligence in medicine: current
trends and future possibilities," Br J Gen Pract. 2018. Available:
https://pmc.ncbi.nlm.nih.gov/articles/PMC5819974/
[3] Talend, "Overcoming Healthcare’s Data Integration Challenges." Available:
https://www.talend.com/resources/data-integration-challenges-healthcare/
[4] Filipe Andrade Bernardi, et al. "Data Quality in Health Research: Integrative Literature Review," J Med
Internet Res, 2023. Available: https://pmc.ncbi.nlm.nih.gov/articles/PMC10646672/
[5] Simbo, "Challenges and Solutions for Ensuring HIPAA Compliance in the Implementation of Artificial
Intelligence in Healthcare." Available: https://www.simbo.ai/blog/challenges-and-solutions-for-
ensuring-hipaa-compliance-in-the-implementation-of-artificial-intelligence-in-healthcare-1580061/
[6] Pew, "How FDA Regulates Artificial Intelligence in Medical Products," 2021. Available:
https://www.pewtrusts.org/en/research-and-analysis/issue-briefs/2021/08/how-fda-regulates-
artificial-intelligence-in-medical-products
[7] Mohamed Khalifa and Mona Albadawy, "AI in diagnostic imaging: Revolutionising accuracy and
efficiency," Computer Methods and Programs in Biomedicine Update, Volume 5, 2024. Available:
https://www.sciencedirect.com/science/article/pii/S2666990024000132
[8] Alexandre Blanco-González, et al., "The Role of AI in Drug Discovery: Challenges, Opportunities, and
Strategies," Pharmaceuticals (Basel), 2023. Available:
https://pmc.ncbi.nlm.nih.gov/articles/PMC10302890/
[9] Nathan Eddy, "Invest in IT Infrastructure to Support AI Adoption in Healthcare," HealthTech, 2023.
Available:
https://healthtechmagazine.net/article/2023/08/invest-it-infrastructure-support-ai-adoption-
healthcare
[10] Sunday Ochella and Mahmood Shafiee, "Performance Metrics for Artificial Intelligence (AI) Algorithms
Adopted in Prognostics and Health Management (PHM) of Mechanical Systems," Journal of Physics:
Conference Series, 2021. Available:
https://iopscience.iop.org/article/10.1088/1742-6596/1828/1/012005/meta
