AI can triage patients and streamline healthcare services in primary care

Important points:

• Machine learning models can effectively triage primary care patients with respiratory symptoms.
• One researcher told Helio that the innovation demonstrates the potential of artificial intelligence in primary care.

Using artificial intelligence for patient triage could streamline healthcare services while tackling issues such as antibiotic resistance, researchers said.

Respiratory symptoms are the most common symptoms seen by primary care physicians. Emil Lars Sigurson, M.D., Ph.D. PCP and Professor at the Department of Family Medicine, University of Iceland, steindall errson,medical doctor, Co-founders and colleagues of an insomnia treatment app wrote: Annals of Family Medicine. Symptoms may indicate a serious illness or may resolve on their own.

Given this extent, triaging patients prior to face-to-face care may reduce physician burden and healthcare costs by providing low-risk patients with alternative means of communication. Artificial intelligence (AI) solutions make this process even faster.

Researchers therefore conducted a retrospective study to understand whether machine learning models could triage patients with respiratory symptoms prior to their visit to a primary care clinic. They found that the technology eliminated the need for chest radiography (CXR) in low-risk patients and reduced the number of their referrals.

Helio spoke with Ellerson and Sigurson to learn about the capabilities of machine learning models, the potential of AI in primary care, and more.

Helio: Can you describe your research? try to find out AI in primary care?

Ellerson and Sigurson: Our study validates the effectiveness of AI models in assessing patients with respiratory symptoms in primary care settings, especially when patients are still at home. Healthcare systems around the world are facing challenges from overuse, leading to increased burnout among healthcare workers. This issue is compounded by the fact that healthcare professionals are spending less and less time with patients. Many patient consultations regarding respiratory illness are for symptoms that do not require clinical evaluation and can be managed with symptomatic treatment alone. However, these patients often undergo unnecessary diagnostic tests and receive antibiotics, which are often inconsistent with clinical guidelines.

As a primary care physician myself, I am particularly interested in research exploring ways to improve service quality and reduce the burden on primary care staff. The use of AI could be a promising approach to addressing these issues by assisting initial patient assessment and triage, potentially reducing unnecessary consultations and interventions.

Helio: What was your discovery and its significance?

Ellerson and Sigurson: This retrospective study suggests that it is possible to classify patients based on symptom severity using only their symptom history prior to seeking help from the health care system. The results indicate that approximately one-third of patients are designated as low risk. This group has the following characteristics:

• Although all their CXRs were negative for pneumonia and tumors, a significant number of these scans revealed incidental findings (incidental tumors) that were often not clinically significant (in our study, all incidental tumors were not significant), leading to increased costs and patient discomfort.
• They have low C-reactive protein values, indicating a high proportion of viral disease.
• This group has a high incidence of International Classification of Diseases (ICD) codes and does not require antibiotics or further medical evaluation. Nevertheless, such treatments are frequent. Interestingly, in the low-risk group, no patients were diagnosed with pneumonia by ICD code or CXR positivity.
• Rates of reassessment within 7 days are low in both primary care settings and emergency departments.

These observations highlight the potential efficiency of patient triage through initial symptom-based assessment aimed at reducing unnecessary medical visits and treatments.

Helio: What are the clinical implications of your research?

Ellerson and Sigurson: At this time, we do not have the ability to determine the severity of a patient’s illness before seeking help from the health system. An AI model that can reliably make this assessment would be more effective in determining where patients should seek help, whether they should stay at home and receive symptomatic care, or whether they should opt for telemedicine consultations over face-to-face consultations. You will be able to guide visit. The COVID-19 pandemic has significantly expanded the use of telemedicine. However, there are no existing guidelines stipulating who is eligible for such consultations. This study shows that physical clinical examinations can be reduced by up to 35%.

Antibiotic resistance is a serious problem worldwide and the most effective way to mitigate this problem is to reduce unnecessary antibiotic use. This study shows that the application of AI can reduce antibiotic use by 25%.

Overuse of imaging is another concern, but our study could reduce imaging referrals by 35% without risking missing diagnoses for serious conditions such as pneumonia. is suggested. Therefore, the introduction of AI in patient triage has the potential to improve the efficiency and accuracy of healthcare delivery.

Helio: Is there anything else you would like to add?

Ellerson and Sigurson: Our research demonstrates the potential of AI in primary care settings. Bringing AI to patient triage can not only streamline healthcare services, but can also make a significant contribution to combating global problems such as antibiotic resistance and overuse of diagnostic resources. This is a great step forward for our continued quest to improve healthcare outcomes while maintaining healthcare system efficiency. But we recognize that this is just the beginning and that more extensive research and testing is needed to refine and optimize the use of AI in medicine.

reference: