Indonesian Political, Business & Finance News

Warning: AI Use Could Send Medical Costs Soaring

| Source: CNBC Indonesia Translated from Indonesian | Finance
Warning: AI Use Could Send Medical Costs Soaring
Image: CNBC Indonesia

Jakarta, CNBC Indonesia — Artificial intelligence (AI) does not always have a positive impact on cost efficiency. Research by the Blue Cross Blue Shield Association (BCBSA) has found that greater use of AI in hospital coding can lead to a rise in hospital bills covered by insurers or patients.

Hospitals commonly use AI to code diagnoses and patient bills. Without AI, diagnosis codes are assigned manually by coders or medical staff, who review patients’ medical records one by one.

AI can make the process much faster, helping coders complete their work more quickly. The technology can read medical records, doctors’ notes and laboratory results, and use this information to identify other conditions or diagnoses that could be included in a bill.

At first glance, AI can make hospital documentation more complete and speed up administrative work. However, the more diagnoses that are recorded, the greater a patient’s complexity may appear in the hospital payment system. As a result, payments from insurers to hospitals can rise.

According to BCBSA’s research, one factor used to determine hospital payments in the United States is the Diagnosis Related Groups (DRG) system. Patients are grouped according to their illnesses and the complexity of their conditions. The more complex a patient’s condition, the higher the payment a hospital may receive.

The issue is that complexity is determined not only by the main illness: additional diagnoses can also move a claim into a more expensive group.

Between 2023 and 2025, around 55,158 additional cases were classified as complex compared with the 2023 baseline. These cases generated approximately US$653 million in additional reimbursement, or about US$11,800 per case.

Overall, BCBSA estimates that these coding changes added around US$942 million to claims costs across the Blue Cross Blue Shield system over two years.

This is not the first time BCBSA has reported such findings. In a March 2026 issue brief, it analysed coding intensity in inpatient care and found that it contributed to higher inpatient costs, despite no corresponding change in the care provided. For example, in maternity care, BCBSA estimated that coding intensity accounted for about US$22 million in additional spending during the period analysed.

To illustrate how AI use in diagnosis could affect hospital costs, CNBC Indonesia Research modelled a hypothetical scenario under Indonesia’s JKN national health insurance scheme.

One of the payment mechanisms used for hospitals under JKN is INA-CBG (Indonesia Case-Based Groups), in which patients are grouped according to their illness, procedures and the severity of their cases. The group assigned determines the package rate paid to the hospital.

According to the Ministry of Health, INA-CBG severity can be affected by secondary diagnoses, such as a patient’s comorbidities or complications. If AI-assisted diagnosis and coding identifies numerous additional diagnoses, those diagnoses could increase the amount paid to the hospital.

For example, an insurer might agree with a hospital on a basic package of Rp20 million per operation and a complex-case package of Rp25 million. If a patient undergoes surgery without any additional diagnosis, the basic rate applies. If an additional diagnosis is recorded, however, the complex package rate is charged — an increase of Rp5 million resulting from the diagnosis identified by AI.

Another example is the analysis of laboratory results. If AI reads a test and finds that a patient has low haemoglobin, possibly due to anaemia, it may recommend the finding to a coder. If the diagnosis is entered and meets coding rules, the complexity profile of the claim could change. Costs could rise further if AI recommends follow-up laboratory tests to investigate a suspected finding in an earlier result.

Mitigation measures

BCBSA’s findings mean hospitals must be cautious in their use of AI to avoid harming insurers or patients. Hospitals and insurers need ways to ensure AI-generated diagnoses accurately reflect a patient’s condition and treatment.

BCBSA found that patients with complex diagnoses received additional care. Its research closely examined major bowel procedures, including operations related to colorectal cancer and diverticular disease.

Between 2023 and 2025, the share of claims in the highest-complexity category, or MCC, rose from 20.2% to 22.7%. By contrast, claims considered non-complex fell from 36.6% to 32.8%. This means more patients were classified as complex, while the care provided did not show a corresponding increase.

The top 25% of hospitals with the fastest growth in coding recorded substantially more complex cases, yet had lower ICU use, lower transfusion rates and unchanged lengths of stay. BCBSA calls this pattern “clinical discordance”: a situation in which diagnoses recorded on claims do not fully correspond to the care provided.

If left unchecked, AI-generated diagnoses could become a new source of medical inflation, an increasingly important issue for the national health insurance industry. As AI use spreads among hospitals in Indonesia, the trends seen in the United States could also emerge here.

For insurers, the findings pose a new challenge. They cannot simply count the procedures performed; they also need to examine the links between diagnoses, treatment, resource use and claim amounts. Under the current health insurance regulatory framework, this role may fall to insurers’ Medical Advisory Boards (MABs).

AI can help hospitals identify and record patients’ diagnoses more effectively. But in a payment system that reimburses more for patients with higher levels of complexity, that capability can also increase healthcare costs.

Insurers’ future challenge, therefore, will not be limited to controlling overutilisation. They will also need to ensure that coding genuinely reflects patients’ conditions. Whereas medical inflation was previously driven mainly by the price and volume of services, in the AI era another factor must not be overlooked: how complex a patient is coded as being.

More from 9 October 2026 →

Cite this article
Source: jawawa.id (https://jawawa.id/newsitem/warning-ai-use-could-send-medical-costs-soaring-1791539538)
Originally published by CNBC Indonesia (https://www.cnbcindonesia.com/research/20261009125804-128-774728/hati-hati-penggunaan-ai-ternyata-bisa-bikin-tarif-medis-melonjak), translated from Indonesian
Print