AI Agents for Infectious Disease Practices: Automating Billing From OPAT Infusion Services to HIV Management and Antimicrobial Stewardship in 2026

Infectious disease is the specialty that gets consulted on everything and compensated for almost none of it. ID physicians provide antimicrobial stewardship recommendations on 20–40 patients per week without billing a single consultation code. They manage complex OPAT infusion regimens where every session involves time-based infusion coding, drug supply charges, and vascular access maintenance — and manual billers routinely miss half the line items. They run HIV panels with quarterly monitoring labs, chronic care management, and prior authorizations for specialty antiretrovirals that each demand precise documentation. The average infectious disease practice leaves $200K–$500K on the table annually because the billing complexity exceeds what manual teams can consistently capture.

ID billing is uniquely fragmented. Revenue comes from inpatient consultations, outpatient chronic disease management, in-office infusion services, travel medicine, transplant monitoring, and stewardship work that spans multiple care settings. Each revenue stream has its own coding rules, documentation requirements, and payer-specific traps. A single OPAT infusion session can generate 4–8 separate line items — initial infusion time, additional time units, sequential drug administrations, drug supply J-codes, waste modifiers, and vascular access device care — and missing any one of them compounds across hundreds of sessions per year.

AI agents built for infectious disease billing capture every billable service across every revenue stream simultaneously — from infusion time calculations and drug supply coding to stewardship consultation capture and chronic disease management enrollment — recovering revenue that practices assumed was unbillable.

OPAT Infusion Services: Where Every Session Is a Multi-Line Billing Event

Outpatient parenteral antibiotic therapy is the highest-revenue-per-visit service most ID practices offer — and the most underbilled. A single OPAT infusion session for a patient receiving vancomycin involves an initial therapeutic infusion (96365 for the first hour), additional infusion time (96366 for each additional hour beyond the first), the drug supply code (J3370 for vancomycin per 500 mg), a waste modifier (-JW) on any unused portion of a single-dose vial, hydration if administered (96360 initial 31 minutes, 96361 each additional hour), and saline flush for line maintenance. When a second antibiotic is administered sequentially — a common scenario with combination therapy for endocarditis, osteomyelitis, or prosthetic joint infections — the sequential infusion adds 96367 for the first hour and 96366 for additional hours of the second drug, plus its own J-code and waste modifier.

Manual billing teams typically capture the primary infusion code and the drug — two of the six to eight billable line items. AI agents read the infusion documentation, calculate precise start-stop times for each drug administered, apply the correct hierarchy of initial, sequential, and concurrent infusion codes, match each antibiotic to its J-code with correct unit calculations, and apply -JW waste modifiers on single-dose vials where the administered dose is less than the vial size.

Infusion Time Calculation and Code Hierarchy

Infusion coding follows a strict hierarchy that determines which drug gets billed as the primary infusion and which is sequential or concurrent:

The primary infusion must be the drug with the highest facility reimbursement rate — a rule that manual billers frequently violate by defaulting to whichever drug was administered first chronologically rather than which pays highest. AI agents compare reimbursement rates across all drugs in a session and assign the highest-paying drug as the 96365 primary infusion, maximizing the initial-hour payment.

Drug Supply J-Codes and Waste Capture

Every OPAT antibiotic has a specific J-code with a specific unit definition, and incorrect unit calculation is one of the most common infusion billing errors:

• Vancomycin (J3370) — per 500 mg; a 1,250 mg dose = 2.5 units, rounded to 3 units with -JW modifier on the 250 mg waste
• Daptomycin (J0878) — per 1 mg; a 500 mg dose = 500 units; single-dose vial at 500 mg means no waste, but weight-based dosing at 420 mg from a 500 mg vial = 80 mg waste with -JW
• Ertapenem (J1335) — per 500 mg; standard 1 g dose = 2 units
• Ceftriaxone (J0696) — per 250 mg; standard 2 g dose = 8 units
• Amphotericin B (J0285) — per 50 mg; weight-based dosing with frequent waste
• Caspofungin (J0637) — per 5 mg; 70 mg loading dose = 14 units, 50 mg maintenance = 10 units

AI agents calculate the exact number of J-code units per administered dose, identify the difference between administered dose and vial size, and apply -JW waste modifiers automatically. For a practice managing 15–40 concurrent OPAT patients, drug supply and waste capture alone recovers $40K–$90K annually — charges that exist in the documentation but never make it to the claim.

HIV Panel Management: Chronic Disease Revenue That Compounds

HIV management generates recurring revenue across multiple billing categories — high-complexity E/M visits, quarterly laboratory monitoring, chronic care management, preventive counseling, and specialty pharmacy prior authorizations — but most ID practices capture only the office visit and labs, leaving the chronic disease management layer entirely unbilled.

E/M and Laboratory Monitoring

HIV patients require quarterly visits with lab monitoring that includes HIV-1 RNA viral load (87536 quantitative), CD4/CD8 lymphocyte panels (86360 absolute CD4 count, 86361 absolute CD8), comprehensive metabolic panels for antiretroviral toxicity monitoring, fasting lipid panels (80061) for protease inhibitor and integrase inhibitor metabolic effects, and hepatitis B surface antigen (87340) and hepatitis C antibody (86803) co-infection screening at baseline. Each visit is a high-complexity E/M encounter (99214 or 99215) because of multi-drug regimen management, resistance pattern monitoring, and co-morbidity assessment.

AI agents ensure modifier -25 is applied when same-day lab draws accompany the E/M visit, verify that the documented complexity supports the billed E/M level, and capture genotypic resistance testing (87901) at treatment initiation and at any point of virologic failure — a $300–$500 test that is frequently omitted from the order set when viral load rebounds above 200 copies/mL.

Chronic Care Management and Remote Patient Monitoring

HIV patients on multi-drug antiretroviral regimens meet the medical complexity threshold for chronic care management (CCM) billing. AI agents identify eligible patients — those with two or more chronic conditions expected to last at least 12 months and placing the patient at significant risk of death, acute exacerbation, or functional decline — and track the non-face-to-face clinical staff time (medication reconciliation, care plan updates, lab result reviews, pharmacy coordination) that qualifies for CCM billing:

• 99490 — First 20 minutes of CCM per calendar month ($42–$64 per patient per month)
• 99491 — First 30 minutes of complex CCM per calendar month ($86–$130 per patient per month)
• 99437 — Each additional 30 minutes of complex CCM ($62–$95 per patient per month)

For a practice managing 500–2,000 HIV patients, enrolling even 30% in CCM at the 99491 rate generates $150K–$450K in annual recurring revenue from services the clinical team is already performing but not documenting in a billable format. AI agents track staff time against each patient, flag patients approaching the monthly threshold, and generate the billing documentation automatically.

PrEP Management

Pre-exposure prophylaxis for HIV prevention is a growing revenue stream that requires its own billing workflow. PrEP visits include preventive medicine counseling (99401 for 15 minutes, 99402 for 30 minutes), renal function monitoring (creatinine clearance for tenofovir-based regimens), STI screening (syphilis, gonorrhea, chlamydia), hepatitis B screening, and quarterly HIV testing to confirm the patient remains negative. AI agents track the PrEP-specific visit cadence — initial visit with full screening panel, then quarterly monitoring — and ensure all required labs are ordered, documented, and billed at each interval. For injectable PrEP (cabotegravir), AI also captures the injection administration code (96372) and drug supply (J0599 per 1 mg) at each bimonthly visit.

Antimicrobial Stewardship: The Revenue You're Giving Away

This is the single largest unbilled revenue category in infectious disease. ID physicians perform antimicrobial stewardship work — reviewing cultures, recommending antibiotic changes, advising on de-escalation protocols, conducting prospective audit and feedback — and bill precisely nothing for it. The stewardship hours go to hospital contract stipends (if the practice is lucky) or pure charity (if they're not).

But billable codes exist for this work, and AI agents capture them.

Interprofessional Consultation Codes

When an ID physician reviews a patient's chart, culture data, and imaging at the request of the treating physician and provides an antibiotic recommendation without seeing the patient face-to-face, this qualifies for interprofessional telephone/internet/electronic health record consultation billing:

The requirements: a treating physician or qualified healthcare professional requests the consultation, the ID physician documents the time spent reviewing records and providing the recommendation, and the consultation results in a written or verbal report back to the requesting provider. That is exactly what stewardship ID physicians do dozens of times per week.

AI agents monitor stewardship documentation — EMR chart review logs, antibiotic recommendation notes, communication records with treating physicians — and match each documented interaction to the appropriate time-based consultation code. A practice performing 25 stewardship consultations per week at an average of 15 minutes each (99447) generates $88K–$130K annually from work that was previously uncompensated.

Inpatient Consultation Coding

For formal inpatient consultations where the ID physician does examine the patient, AI ensures correct coding based on 2021 E/M guidelines — time-based or medical decision-making complexity. Initial hospital consultations (99221–99223) and subsequent hospital care (99231–99233) are selected based on documented total time on the date of encounter, including chart review, face-to-face examination, order entry, and documentation. AI captures prolonged services (99418 for each additional 15 minutes beyond the typical time for the billed code) when complex cases — multi-drug resistant organisms, sepsis with source control questions, transplant infections — require extended physician involvement.

Hepatitis C Treatment: Cure-Based Billing With Prior Authorization Gatekeeping

HCV treatment is a finite, cure-oriented treatment course — typically 8–12 weeks of direct-acting antivirals — but the billing and prior authorization complexity rivals any chronic disease. Payers require extensive documentation before approving DAA therapy, and incomplete PA submissions are the primary reason for the 30–40% initial denial rate that delays treatment initiation.

AI agents compile the complete PA package automatically:

1. HCV genotype testing (87902) — AI verifies genotype is documented and includes it in the PA
2. Fibrosis staging — FibroScan transient elastography results or FibroSure/FibroTest biomarker panel, with kPa values or METAVIR score documented
3. Prior treatment history — Previous HCV treatment attempts, regimens used, and outcome (relapse, non-response, treatment-naive)
4. Contraindication screening — Drug interaction review against current medication list, hepatitis B co-infection status, renal function for sofosbuvir-based regimens
5. Prescriber specialty verification — Some payers require the prescribing physician to be a gastroenterologist, hepatologist, or infectious disease specialist

AI tracks the treatment timeline once DAA therapy starts — ensuring week 4 viral load monitoring is ordered, capturing the end-of-treatment visit, and scheduling the SVR12 (sustained virologic response at 12 weeks post-treatment) confirmatory viral load that proves cure. Missing the SVR12 window doesn't just lose the billing for that visit — it loses the documented cure that justifies the $20K–$90K drug cost to the payer and prevents re-treatment authorization disputes.

Travel Medicine: High-Volume Vaccine Revenue With Complex Charge Capture

Travel medicine visits are among the most charge-dense encounters in infectious disease — a single pre-travel consultation can generate 8–12 separate line items — yet they're frequently billed as a single office visit because the vaccine administration coding complexity overwhelms manual billing workflows.

Every vaccine requires two separate charges: the product code (the vaccine itself) and the administration code (the act of giving it). The first injection in a visit bills 90471 ($28–$42), and each additional injection bills 90472 ($14–$21). A traveler to sub-Saharan Africa receiving yellow fever (90717), typhoid (90691), hepatitis A (90632), hepatitis B (90746), and meningococcal (90734) vaccines generates five product codes plus five administration codes: 90471 × 1 + 90472 × 4 = five administration charges alone worth $84–$126.

AI agents match destination-specific CDC recommendations to the patient's immunization history, identify which vaccines are indicated versus already documented, generate both product and administration codes for every vaccine given, and capture the pre-travel counseling visit as preventive medicine counseling (99401–99404) based on documented counseling time. For peak travel season (April–August), a busy travel clinic seeing 30–60 patients per week recovers $1,500–$3,000 per week in previously missed administration codes and counseling charges.

Transplant Infectious Disease: Lifelong Monitoring Revenue

Post-transplant infectious disease monitoring generates recurring revenue for years — CMV viremia surveillance (87497 CMV DNA quantitative PCR), EBV monitoring for PTLD risk (87799), BK virus monitoring in renal transplant recipients (87799), Pneumocystis and toxoplasmosis prophylaxis management, and the complex medication interactions between immunosuppressants and antimicrobials that require ID involvement at every regimen change.

AI agents track transplant patient panels with automated monitoring schedules — weekly CMV PCR for the first 3 months, biweekly for months 3–6, monthly for months 6–12 — and ensure each monitoring visit captures the E/M encounter, quantitative PCR lab codes, and medication management documentation. When CMV viremia is detected and antiviral treatment (valganciclovir or IV ganciclovir) is initiated, AI captures the treatment E/M visits, weekly follow-up viral load monitoring, and drug supply charges for IV ganciclovir (J1570 per 500 mg) when administered in the infusion center.

For practices managing 100–300 transplant patients, monitoring revenue alone generates $60K–$150K annually — revenue that's consistent, recurring, and high-margin because the monitoring infrastructure already exists.

The Full Infectious Disease Revenue Cycle — Automated

AI agents don't optimize one revenue stream — they capture all of them simultaneously across the fragmented ID billing landscape:

• OPAT infusion services — Time-based infusion coding hierarchy, drug supply J-code unit calculation, -JW waste capture, sequential and concurrent infusion coding, vascular access maintenance
• HIV panel management — High-complexity E/M coding, quarterly lab monitoring capture, CCM/complex CCM enrollment and time tracking, PrEP management workflows, resistance testing triggers
• Antimicrobial stewardship — Interprofessional consultation code capture (99446–99449), inpatient consultation coding, prolonged service documentation
• Hepatitis C treatment — Automated PA compilation, treatment timeline tracking, SVR12 scheduling, genotype and fibrosis documentation
• Travel medicine — Vaccine product and administration code pairing, destination-based recommendation matching, preventive counseling time capture
• Transplant monitoring — Automated surveillance schedules, quantitative PCR coding, treatment episode capture, recurring visit billing

For infectious disease practices running 3–8 physicians, the combined impact is $200K–$500K in annual revenue recovery — the majority from services that were already being performed but never made it onto a claim.