Blab AI
Brazilian Woman Dies After AI Bed-Allocation System Delays ICU Transfer
A 32‑year‑old woman from São João Nepomuceno, Brazil, died after a state‑run artificial‑intelligence (AI) system that assigns hospital beds delayed her transfer to an intensive care unit (ICU) for five days. The incident highlights the risks of relying on automated scoring for critical patient triage.
Rebeca Cardoso Tenente Molina was admitted to a local hospital in São João Nepomuceno after seeking treatment for gallstones. As her condition worsened, she required an ICU bed in the state of Oliveira, a facility 186 miles away. According to the Brazilian news outlet MG1, Molina’s family pursued emergency legal action to secure a transfer, but the AI‑based bed‑allocation system, known as Core‑MG, maintained her score at 6.8 while her clinical status was described by family members as severe.
Sâmela Cardoso Tenente Furtado, Molina’s sister and a family lawyer, told MG1 that Core‑MG’s automatic scoring algorithm assigned a lower priority to Molina than her condition warranted. She said the system would not accept an increase in her severity level, even when test data indicated a deterioration. Furtado added that doctors had lost the ability to override the AI’s assessment, stating that the system “accepted her as a 6.8 when she should have been a 10.”
Core‑MG was launched on 19 May by the Minas Gerais State Health Secretariat. The system uses machine‑learning models to generate a bed‑map that is updated three times a day, according to Deputy Secretary of Health Poliana Cardoso Lopes. Lopes said the tool would give “better control over the process and generate better data on the clinical condition and needs of each person waiting for a bed.”
In response to Molina’s death, the state health department clarified that patient transfers are still determined by the availability of beds that match a patient’s clinical needs. The department stated that Core‑MG has not fundamentally altered the protocol for transferring patients to other facilities.
The case raises questions about the safety and transparency of AI systems used for patient triage. While AI can process large volumes of data quickly, the incident shows that automated scoring can delay critical care when the algorithm’s thresholds or data inputs do not align with real‑time clinical observations. The family’s legal action underscores the need for mechanisms that allow medical staff to override or audit AI decisions in urgent situations.
Minas Gerais officials have not yet announced any changes to Core‑MG’s scoring methodology or governance structure. The state’s health department has said it will continue to monitor the system’s performance, but no specific timeline for adjustments has been released.
The incident adds to a growing body of evidence that AI tools in healthcare must be accompanied by robust oversight, clear accountability, and safeguards that preserve clinician judgment, especially when patient outcomes are at stake.
The broader implications for AI adoption in hospital management include the importance of transparent algorithms, real‑time validation against clinical data, and the inclusion of human‑in‑the‑loop controls. As more jurisdictions deploy similar systems, regulators and health authorities will need to establish standards for algorithmic risk assessment and emergency override procedures.
Until further changes are announced, patients in Minas Gerais will continue to rely on Core‑MG for bed allocation, but the system’s role in critical care decisions remains under scrutiny. The state’s health department has pledged to review the incident and assess whether additional safeguards are necessary to prevent future delays in life‑saving care.
Rebeca Cardoso Tenente Molina was admitted to a local hospital in São João Nepomuceno after seeking treatment for gallstones. As her condition worsened, she required an ICU bed in the state of Oliveira, a facility 186 miles away. According to the Brazilian news outlet MG1, Molina’s family pursued emergency legal action to secure a transfer, but the AI‑based bed‑allocation system, known as Core‑MG, maintained her score at 6.8 while her clinical status was described by family members as severe.
Sâmela Cardoso Tenente Furtado, Molina’s sister and a family lawyer, told MG1 that Core‑MG’s automatic scoring algorithm assigned a lower priority to Molina than her condition warranted. She said the system would not accept an increase in her severity level, even when test data indicated a deterioration. Furtado added that doctors had lost the ability to override the AI’s assessment, stating that the system “accepted her as a 6.8 when she should have been a 10.”
Core‑MG was launched on 19 May by the Minas Gerais State Health Secretariat. The system uses machine‑learning models to generate a bed‑map that is updated three times a day, according to Deputy Secretary of Health Poliana Cardoso Lopes. Lopes said the tool would give “better control over the process and generate better data on the clinical condition and needs of each person waiting for a bed.”
In response to Molina’s death, the state health department clarified that patient transfers are still determined by the availability of beds that match a patient’s clinical needs. The department stated that Core‑MG has not fundamentally altered the protocol for transferring patients to other facilities.
The case raises questions about the safety and transparency of AI systems used for patient triage. While AI can process large volumes of data quickly, the incident shows that automated scoring can delay critical care when the algorithm’s thresholds or data inputs do not align with real‑time clinical observations. The family’s legal action underscores the need for mechanisms that allow medical staff to override or audit AI decisions in urgent situations.
Minas Gerais officials have not yet announced any changes to Core‑MG’s scoring methodology or governance structure. The state’s health department has said it will continue to monitor the system’s performance, but no specific timeline for adjustments has been released.
The incident adds to a growing body of evidence that AI tools in healthcare must be accompanied by robust oversight, clear accountability, and safeguards that preserve clinician judgment, especially when patient outcomes are at stake.
The broader implications for AI adoption in hospital management include the importance of transparent algorithms, real‑time validation against clinical data, and the inclusion of human‑in‑the‑loop controls. As more jurisdictions deploy similar systems, regulators and health authorities will need to establish standards for algorithmic risk assessment and emergency override procedures.
Until further changes are announced, patients in Minas Gerais will continue to rely on Core‑MG for bed allocation, but the system’s role in critical care decisions remains under scrutiny. The state’s health department has pledged to review the incident and assess whether additional safeguards are necessary to prevent future delays in life‑saving care.
