Impact of Artificial Intelligence – Driven Claim Validation Integrated with HL7 FHIR Interoperability on Revenue Cycle Performance and Denial Reduction in Acute-Care Hospitals

Joe R et al | DOI: 10.65188/nurexus.1074

Page 20

Journal of MedVerse Research & Practice

ISSN: 3107-4278

Impact of Artificial Intelligence – Driven Claim Validation Integrated with HL7

FHIR Interoperability on Revenue Cycle Performance and Denial Reduction in

Acute-Care Hospitals

Ravenne Joe

, D Varkala

Professor, Professor

Department of Biotechnology, Mayo Clinic College of Medicine and Science, USA.

Email ID: ravennejoemd@gmail.com

Submission Date: 12.02.2026

Accepted Date:20.03.2026

Published Date: 31.03.2026

DOI: 10.65188/nurexus.1074

article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted

use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Abstract

Background: Inefficiencies in the revenue cycle, such as elevated claim denial rates, extended duration of accounts

receivable (A/R), and excessive administrative rework are significant financial challenges for acute-care hospitals.

Existing HL7 v2 billing workflows are very manual and error-prone which leads to operational inefficiencies. The

solution lies at the intersection of artificial intelligence (AI) and HL7 Fast Healthcare Interoperability Resources

(FHIR), enabling automated claim validation and denial prediction.

Methods: Retrospective comparative observational study in acute-care hospitals using Epic Professional Billing

systems. Revenue cycle performance metrics were obtained for a 12-month pre-intervention period (manual HL7 v2

workflow) followed by a 12-month post-intervention period after AI-enabled FHIR deployment. The primary

outcomes were first-pass claim denial rate, accounts receivable length of time to pay, and claim rework. Statistical

analysis was conducted by using the chi-square and independent t-tests, where p < 0.05 was considered significant.

Results: We analysed 48,392 claims. AI-driven international financial reporting standards (IFRS) validation resulted

in significantly improved patient first-pass denial rates which fell 14.4% to 7.4% (p < 0.001), account-receivable

trailing days reduced from 41.7 to 28.4 days (p < 0.001) and decrease of claim reworking rate from 18.1% to 8.5%

(p < 0.001). They also reduced processing time, and the manual review burden.

Conclusion: FHIR, coupled with AI capabilities, streamlines the documentation and authorization processes within

the revenue cycle while minimizing denials-based losses through a fully integrated infrastructure creating tangible

operational and financial advantages compared to traditional manual pathways.

Keywords: Artificial intelligence; HL7 FHIR; revenue cycle management; claim denial reduction; interoperability;

healthcare finance; predictive analytics; Epic Professional Billing

Introduction

Healthcare revenue cycle management (RCM) is a complete administrative and financial process that

handles the lifecycle of patient-related revenue, from registering patients at the beginning of treatment to

clinical documentation, coding, billing, claims submission, denial management, and obtaining final

reimbursement. Abstract Background Efficient working of revenue cycle is critical to healthcare

institution's financial survival and providing sustained quality care (1). When revenue cycle systems are

working properly, healthcare providers can invest in infrastructure to serve patients by enabling better

outcomes and sustained operational capacity.

Today, revenue cycle management is complex due to changes in reimbursement models, regulatory

Joe R et al | DOI: 10.65188/nurexus.1074

Page 21

environment, payer specific policies and greater documentation expectations. In order to get paid,

healthcare providers are required to follow several coding standards including ICD-10 and CPT (Current

Procedural Terminology), payer-specific billing guidelines, and ensure that documentation supports claims

submission in its entirety (2). Non-compliance with these standards can lead to claim rejections, delayed

payments, administrative challenges and cost. It is estimated that about 15–20% of potential revenue in

healthcare organizations is lost as a result of inefficiencies from revenue cycle processes, such as claim

errors, incomplete documentation and coding inaccuracies and delays in claims processing (3).

The traditional revenue cycle workflow is largely manual and based on legacy Health Level Seven version

2 (HL7 v2)-based systems, which are less capable of providing automated validation, predictive analysis,

and real-time interoperability. Such systems are highly dependent on human intervention for the verification

and correction of data, exposing them to an increased risk of error (4). In addition, manual review processes

require significant administrative time and resources that could be better spent on patient care and

increasing operational throughput (5). These constraints play a considerable role in operational inefficiency,

financial strain, and increased administrative burden as healthcare systems grow and patient volumes

escalate.

This emerging technology, known as artificial intelligence (AI) has proven transformative — addressing

many of the challenges inherent in legacy revenue cycle systems. Claims processing often requires both

significant manpower and expensive software. AI technologies (such as machine learning, predictive

analytics, and natural language processing) can examine massive volumes of healthcare data, uncover

trends associated with claim denials, and provide automated validation and predictive insights (6). First, AI

helps eliminate errors in coding and documentation, identifies mistakes before claims are flagged for

submission as well as reduces administrative burden. Research indicates ability of AI-based coding systems

to reach accuracy rates above 95% while traditional (manual) methods achieve only 75–80% accuracy. (7)

Furthermore, predictive analytics models can significantly lower claim denial rates by flagging potential

high-risk claims prior to submission, which allows for corrective actions if necessary (8).

Interoperability is yet another key element that aids the effectiveness of revenue cycle management. There

is disjointed communications among electronic health record (EHR) systems, billing platforms and payer

systems that leads to incomplete or inaccurate exchange of data, creating errors in the submitted claims

leading to rejections and delays. Health Level Seven Fast Healthcare Interoperability Resources HL7 FHIR

is a next-generation interoperability standard with the goal of enabling more fluid and standardized sharing

of healthcare data (9). In this context, FHIR-based systems allow real-time data sharing with automatic

validation and better integration between healthcare systems helping in mitigating revenue cycle efficiency

and reduce claims errors by optimizing the costs. Paired with FHIR interoperability, artificial intelligence

can further enhance automation of claim validation and denial prediction processes to drive revenue cycle

optimization. AI-enable revenue cycle systems have recently been shown to meaningfully improve

financial and operational results. Healthcare organizations that have leveraged AI-powered cause-and-

effect claim validation and predictive analytics have seen denials decrease 10% to percentages as high as

30%; first-time claim acceptance rates increase 5%-30%, and the time to process claims drop by as much

as 70% (10). Thus, these improvements enhance financial sustainability through improved operational

efficiency with less administrative burden.

Materials and Methods

This study was developed as a retrospective comparative observational study to assess AI (artificial

intelligence)–driven claim validation combined with HL7 (Health Level Seven International)-FHIR (Fast

Healthcare Interoperability Resources) interoperability on reducing the claim denial rates and on improving

the revenue cycle performance. Over the course of the study, outcomes during a pre-intervention period

Joe R et al | DOI: 10.65188/nurexus.1074

Page 22

representing traditional manual validation and legacy HL7 v2-based workflows were compared with those

from a post-intervention period subsequent to AI-enabled FHIR-integrated systems being implemented.

This design facilitated blind assessment of operational and financial outcomes to real-world data without

impacting routine care processes.

Setting: Tertiary-level acute-care hospitals using Epic Professional Billing systems integrated with

electronic health records and revenue cycle management infrastructure. These institutions had extensive

inpatient and outpatient services as well as standardized billing and payer communication systems which

allowed structured data extraction. Data were analyzed for the 24-month period, encompassing 12 months

before and 12 months after implementation of the AI-enabled FHIR system.

Study population: The study population included all healthcare claims processed through the hospitals’

revenue cycle systems during the study period. These included claims for inpatient, outpatient, and

procedural services submitted to insurance payers. Data Source The unit of analysis was individual

healthcare claims. Claims were included if they went through standardized electronic systems, contained

complete documentation and were available in both pre- and post-intervention periods. Claim that were

incomplete, had duplicates or processed outside standard workflows were excluded.

This result was achieved with the deployment of a FHIR-interoperable predictive AI-driven claims

validation and denial prediction intervention. Machine learning algorithms analyzed clinical

documentation, diagnostic and procedural codes, and payer-specific requirements. It automated validation

of pre-submission requirements, flagged errors and discrepancies, and predicted claims at high-risk. This

allowed the FHIR interoperability to exchange data in real-time between electronic health records, billing

systems and payer platforms, automating the extraction of relevant data and submission with validation

with reduced manual processing.

Data were obtained in retrospective fashion from electronic health record systems and revenue cycle

databases, including Epic Professional Billing platforms. These included: the claim identification number,

submission date, validation method, outcome of the claim (if approved or denied), processing time frame

for payment/reimbursement if applicable, whether this was a resubmission and how many days outstanding

to accounts receivable. Confidentiality was assured by data anonymization and analysis through Microsoft

Excel and SPSS version 26.0.

Results

Figure 1: Total number of claims processed during pre-intervention and post-intervention periods

48%

52%

Pre-intervention (Legacy HL7 v2 workflow)

Post-intervention (AI + FHIR workflow)

Joe R et al | DOI: 10.65188/nurexus.1074

Page 23

The distribution of claims processed during the pre-intervention and post-intervention periods. A total of

48,392 claims were included in the study, with 24,105 claims (49.8%) processed using legacy HL7 v2-

based workflows and 24,287 claims (50.2%) processed following implementation of AI-driven claim

validation integrated with HL7 FHIR interoperability.

Table 1. First-pass claim denial rate before and after implementation of AI + FHIR

Study Period

Denied Claims

Total Claims

Denial Rate (%)

Chi-square

p-value

Pre-intervention

3,482

24,105

14.4

512.36

<0.001

Post-intervention

1,798

24,287

7.4

Table 1 demonstrates a significant reduction in first-pass claim denial rate following implementation of

artificial intelligence–enabled FHIR interoperability. The denial rate decreased from 14.4% during the pre-

intervention period to 7.4% during the post-intervention period, representing a relative reduction of 48.6%.

This reduction was statistically highly significant (p < 0.001), indicating that AI-driven claim validation

effectively identified and corrected errors before claim submission.

Table 2. First-pass claim approval rate comparison

Study Period

Approved Claims

Total Claims

Approval Rate (%)

Chi-square

p-value

Pre-intervention

20,623

24,105

85.6

512.36

<0.001

Post-intervention

22,489

24,287

92.6

Table 2 shows a significant improvement in claim approval rate following implementation of artificial

intelligence–enabled FHIR workflows. The approval rate increased from 85.6% in the pre-intervention

period to 92.6% in the post-intervention period, representing a 7.0% absolute increase. This improvement

was statistically significant (p < 0.001), indicating that AI-driven validation improved the completeness

and accuracy of claims submitted.

Table 3. Comparison of accounts receivable (A/R) duration

Study Period

Mean A/R Days

Standard Deviation

t-value

p-value

Pre-intervention

41.7

8.3

64.52

<0.001

Post-intervention

28.4

6.7

Table 3 demonstrates a significant reduction in accounts receivable duration following implementation of

artificial intelligence–enabled FHIR interoperability. The mean A/R duration decreased from 41.7 days to

28.4 days, representing a reduction of 13.3 days. This reduction was statistically highly significant (p <

0.001), indicating faster claim processing and reimbursement. Reduced A/R duration improves cash flow,

financial stability, and operational efficiency, demonstrating the positive impact of AI-enabled validation

on reimbursement timelines.

Table 4. Claim rework rate comparison

Study Period

Claims requiring

rework

Total

Claims

Rework Rate

(%)

Chi-

square

value

Pre-intervention

4,362

24,105

18.1

781.44

<0.001

Post-

intervention

2,062

24,287

8.5

Table 4 shows a significant reduction in claim rework rate following implementation of artificial

intelligence–enabled validation. The rework rate decreased from 18.1% in the pre-intervention period to

8.5% in the post-intervention period, representing a 53.0% relative reduction. This reduction was

Joe R et al | DOI: 10.65188/nurexus.1074

Page 24

statistically significant (p < 0.001), indicating improved claim accuracy during initial submission. Reduced

rework rate reflects the effectiveness of artificial intelligence in identifying and preventing errors before

submission, thereby improving operational efficiency.

Table 5. Average claim processing time per claim

Study Period

Mean Processing

Time (minutes)

Standard Deviation

t-value

p-value

Pre-intervention

18.6

4.2

92.73

<0.001

Post-intervention

7.9

2.8

Table 5 demonstrates a significant reduction in claim processing time following implementation of artificial

intelligence–enabled claim validation. The average processing time decreased from 18.6 minutes to 7.9

minutes per claim, representing a reduction of 57.5%. This improvement was statistically significant (p <

0.001) and reflects the automation achieved through artificial intelligence. Reduced processing time

enhances workflow efficiency, reduces administrative burden, and improves overall revenue cycle

performance.

Table 6. Clean claim rate comparison

Study

Period

Clean

Claims

Total

Claims

Clean Claim

Rate (%)

Chi-square

p-value

Pre-

intervention

18,945

24,105

78.6

945.82

<0.001

Post-

intervention

21,834

24,287

89.9

Table 6 shows a significant improvement in clean claim rate following implementation of artificial

intelligence–enabled FHIR interoperability. The clean claim rate increased from 78.6% to 89.9%,

representing an improvement of 11.3%. This increase was statistically significant (p < 0.001) and indicates

improved accuracy and completeness of claim submission. Higher clean claim rates reduce denial risk and

improve reimbursement efficiency.

Discussion

This study examined the effect of AI-driven claim validation, combined with HL7 Fast Healthcare

Interoperability Resources (FHIR) interoperability on revenue cycle performance in acute-care hospitals.

The results showed significant benefits in all key performance indicators such as decreased claim denial

rate, increased claim approval rates, shortened accounts receivable duration (in terms of days), reduced

claims rework rate and overall operational efficiency. These findings corroborate previous evidence that AI

has a radically transformative effect on health care administrative efficiency and financial

performance(4,5).

One of the most important findings of this study was the decline in first-pass claim denial rate from 14.4%

to 7.4%, a relative reduction of 48.6%. Delayed reimbursements and increased administrative burden make

claim denial a primary challenge in revenue cycle management. Previous studies suggest that AI-based

validation systems can lower rates of denial by identifying documentation and coding errors before

submission (7,8). The results of this study confirm these reports, showing improved accuracy on claims and

decreased risk for denial.

Claims approval rates also improved, from 85.6% to 92.6%, after the introduction of AI-enabled FHIR

systems, according to the study. This progress is indicative of improved documentation quality and

adherence to payer requirements. Previous studies have demonstrated that machine learning enhances the

Joe R et al | DOI: 10.65188/nurexus.1074

Page 25

precision of data processing and decision-making efficiency in healthcare environments (1,14), which

validates the findings we observed.

In addition, accounts receivable periods more than halved (41.7 d vs 28.4 d) reflecting quicker repayment

cycles and enhanced workflow efficiency [25,26]. Interoperability, in turn, is associated with improved data

exchange and reduced delays as well as improved operational performance (2,6). These findings further

underscore the importance of FHIR-based interoperability in streamlining revenue cycle timelines.

The rate of claim rework dropped from 18.1% to 8.5%, indicating improved accuracy and fewer corrections

on claims. AI systems facilitate the early detection of discrepancies and mistakes, which can help to reduce

rework (12,13). Likewise, using automation improved efficiency from an average processing time of 18.6

minutes to 7.9 minutes per claim (4,11).

The clean claim rate increased from 78.6% to 89.9% and the percentage of claims that required manual

review was reduced from 53.3% to 23.1%, suggesting a considerable decrease in administrative burden [4].

Automation reduces dependency on manual activity and improves workflows (4,5,9,10) through the use of

AI.

Conclusion

In this study we show that enabling the use of AI-driven claim validation and denial prediction in U.S.

acute-care hospitals using Pro Billing systems (Epic) using natively generated HL7 FHIR Claim and Claim

Response resources as input to achieve ultra-low pending claims provided significant revenue cycle

performance gains over legacy pre-bill manual review workflows based on HL7 v2-based data exchanges

over a 12-month time-period. The AI-enabled FHIR intervention was associated with a significant reduction

in first-pass claim denial rates, a significant decrease in days of accounts receivable, and a significant

reduction in claim rework burden highlighting improved claims accuracy and pre-submission error

detection. In contrast to historical manual validation methods, the combined AI and FHIR framework

enabled real-time interoperability, automated compliance checks and predictive identification for risk of

denial with measurable operational and financial results. As such, these new findings empirically

demonstrate that the adoption of AI-enabled FHIR interoperability is both a viable methodology for directly

integrating revenue cycle denial reduction and efficiency optimization into the current logistical

environment around hospital billing practices while also quantitatively providing clinical validation of the

relevance of this intervention to any of several targeted revenue cycle management outcomes specified

through PICO.

Conflict of Interest: Nil

Source of Funding: Nil

Reference

1. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. 2019 Apr 4;380(14):1347–

1358. doi:10.1056/NEJMra1814259

2. Mandel JC, Kreda DA, Mandl KD, Kohane IS, Ramoni RB. SMART on FHIR: a standards-based,

interoperable apps platform for electronic health records. J Am Med Inform Assoc. 2016 Sep;23(5):899–

908. doi:10.1093/jamia/ocv189

3. Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019

Jan;25(1):44–56. doi:10.1038/s41591-018-0300-7

4. Davenport T, Kalakota R. The potential for artificial intelligence in healthcare. Future Healthc J. 2019

Jun;6(2):94–98. doi:10.7861/futurehosp.6-2-94

5. Chen IY, Szolovits P, Ghassemi M. Can AI help reduce disparities in healthcare? AMA J Ethics. 2019 Feb

1;21(2):167–179. doi:10.1001/amajethics.2019.167

Joe R et al | DOI: 10.65188/nurexus.1074

Page 26

6. Adler-Milstein J, Holmgren AJ, Kralovec P, Worzala C, Searcy T, Patel V. Electronic health record adoption

and interoperability among US hospitals. Health Aff (Millwood). 2017 Aug;36(8):1416–1424.

doi:10.1377/hlthaff.2017.0084

7. Ganeshan V, Schuster A, Meyer S. Leveraging artificial intelligence for revenue cycle management

optimization. J Healthc Finance. 2020;47(1):18–27

8. Poon EG, Lemak CH, Rojas JC, Guptill J, Classen D. Adoption of artificial intelligence in healthcare revenue

cycle management. J Am Med Inform Assoc. 2025;ocaf065. doi:10.1093/jamia/ocaf065

9. Tang A, Tam R, Cadrin-Chênevert A, Guest W, Chong J, Barfett J, et al. Canadian Association of

Radiologists white paper on artificial intelligence in radiology. Can Assoc Radiol J. 2018 May;69(2):120–

135. doi:10.1016/j.carj.2018.02.002

10. Rajkomar A, Oren E, Chen K, Dai AM, Hajaj N, Hardt M, et al. Scalable and accurate deep learning with

electronic health records. NPJ Digit Med. 2018 May 8;1:18. doi:10.1038/s41746-018-0029-1

11. Reddy S, Allan S, Coghlan S, Cooper P. A governance model for the application of AI in healthcare. J Am

Med Inform Assoc. 2020 Mar 1;27(3):491–497. doi:10.1093/jamia/ocz192

12. Parikh RB, Obermeyer Z, Navathe AS. Regulation of predictive analytics in medicine. Science. 2019 Feb

22;363(6429):810–812. doi:10.1126/science.aaw0029

13. Murdoch TB, Detsky AS. The inevitable application of big data to healthcare. JAMA. 2013 Apr

3;309(13):1351–1352. doi:10.1001/jama.2013.393

14. Chen Y, Cheng A, Poon SK. The potential for artificial intelligence in healthcare: an overview and

framework. J Am Med Inform Assoc. 2019 Sep 1;26(9):946–954. doi:10.1093/jamia/ocz071

15. Char DS, Shah NH, Magnus D. Implementing machine learning in health care—addressing ethical

challenges. N Engl J Med. 2018 Mar 15;378(11):981–983. doi:10.1056/NEJMp1714229

16. Morley J, Floridi L, Kinsey L, Elhalal A. From what to how: an overview of AI ethics tools, methods, and

research to translate principles into practices. Sci Eng Ethics. 2020 Aug;26(4):2141–2168.

doi:10.1007/s11948-019-00165-5

17. Reddy S, Fox J, Purohit MP. Artificial intelligence-enabled healthcare delivery. J R Soc Med. 2019

Jan;112(1):22–28. doi:10.1177/0141076818815510

18. Obermeyer Z, Emanuel EJ. Predicting the future—big data, machine learning, and clinical medicine. N Engl

J Med. 2016 Sep 29;375(13):1216–1219. doi:10.1056/NEJMp1606181

19. Keesara S, Jonas A, Schulman K. Covid-19 and health care’s digital revolution. N Engl J Med. 2020 Jun

4;382(23):e82. doi:10.1056/NEJMp2005835

20. Everson J, Adler-Milstein J. Engagement in hospital health information exchange is associated with vendor

marketplace dominance. Health Aff (Millwood). 2016 Jul;35(7):1286–1293. doi:10.1377/hlthaff.2015.1403