SINGAPORE: Some patients stricken with pneumonia may soon be prescribed antibiotics with the help of artificial intelligence.
The information these patients provide on their symptoms will be fed into a system, which will use AI to deduce whether the patient is suffering from a viral or bacterial form of the illness.
The system will then decide whether antibiotics - which are used only in cases of bacterial infection - are needed. It will also recommend the specific antibiotic needed, as well as the dose based on the patient's clinical data.
Singapore General Hospital (SGH) developed the AI solution with IT firm DXC Technology and national health tech agency Synapxe.
Pneumonia - a serious inflammatory condition of the lungs - is a leading cause of death worldwide and the fourth biggest cause of hospitalisation at local hospitals.
The research team chose to focus first on pneumonia as it can become very severe very fast, said team member Piotr Chlebicki, a senior consultant in SGH’s Department of Infectious Diseases.
The power of the new tech-driven method lies in its ability to comb through vast amounts of information to detect patterns suggestive of infection, said Dr Chlebicki.
“That will tell us that this patient may be at risk of developing infection and this patient may need antibiotics,” he added.
Data from about 8,000 patients seen by SGH between 2019 and 2020 was used to build the AI pneumonia model for a pilot study.
This data included X-rays, clinical symptoms, periodic vital signs, and trends of common body responses to infection.
The information was then validated against another 2,000 cases in 2023.
The pilot study found that the system, called Augmented Intelligence in Infectious Diseases (AI2D), accurately determined if antibiotics were necessary upon a patient's first diagnosis in nine out of 10 cases, SGH said.
The use of such technology is expected to reduce over-prescription of antibiotics as part of efforts to fight the growing problem of drug-resistant bugs.
Dr Chlebicki noted that it is often hard to tell definitively that patients will benefit from antibiotics based on clinical assessment, patient-specific factors, or condition severity alone.
However, dire complications can arise from belated prescriptions of antibiotics for those who genuinely need them.
The current standard is that doctors prescribe antibiotics on suspicion of pneumonia without waiting for confirmation, as some test results may take up to two days, he said.
“We cannot wait that long. If we wait without starting treatment and make a mistake, the patient can possibly get sicker, so the initial day or two in treatment of pneumonia are very crucial,” noted Dr Chlebicki.
“Yet, misusing antibiotics contributes to antibiotic resistance, posing challenges for future infection treatment. A tool like AI2D will be very helpful to guide doctors’ decisions before lab results are available.”
The cautious attitude towards pneumonia treatment means many patients who were initially suspected to have the infection but later confirmed not to have one, would, in retrospect, have been prescribed antibiotics unnecessarily, Dr Chlebicki noted.
The pilot study revealed that last year, almost 40 per cent of antibiotics prescribed to patients to treat pneumonia at the onset may not have been necessary.
Going forward, the team, which is led by SGH’s Division of Pharmacy, will measure the system’s effectiveness in safely reducing antibiotic use in a comparative study involving 200 inpatients.
These patients will be randomly assigned to doctors who could use AI2D to figure out whether to give them antibiotics.
Antimicrobial resistance (AMR) - when drugs do not work against viruses, bacteria and fungi - makes infections harder to treat and increases the risks of other medical procedures and treatments, according to the World Health Organization.
These include surgery, caesarean sections and chemotherapy for cancer.
One person is projected to die every three seconds from AMR by 2050 if the global problem is not addressed.
Associate Professor Andrea Kwa, deputy director of pharmacy (research and innovation) at SGH, described AMR as a “silent pandemic” that is “coming up fast and fierce”.
“That led us to think of a solution that will quickly help our doctors decide whether this patient needs antibiotics for his or her pneumonia,” said Assoc Prof Kwa, who led the project.
The process can help doctors save up to 20 minutes per case, she said.
“If we don't have to prescribe antibiotics, we are able to retard the emergence of further resistance,” she added.
While the model undergoes further development, the team hopes it can eventually be used for other common conditions like urinary tract infections, and be expanded to other healthcare institutions.
Continue reading...
The information these patients provide on their symptoms will be fed into a system, which will use AI to deduce whether the patient is suffering from a viral or bacterial form of the illness.
The system will then decide whether antibiotics - which are used only in cases of bacterial infection - are needed. It will also recommend the specific antibiotic needed, as well as the dose based on the patient's clinical data.
Singapore General Hospital (SGH) developed the AI solution with IT firm DXC Technology and national health tech agency Synapxe.
Pneumonia - a serious inflammatory condition of the lungs - is a leading cause of death worldwide and the fourth biggest cause of hospitalisation at local hospitals.
The research team chose to focus first on pneumonia as it can become very severe very fast, said team member Piotr Chlebicki, a senior consultant in SGH’s Department of Infectious Diseases.
The power of the new tech-driven method lies in its ability to comb through vast amounts of information to detect patterns suggestive of infection, said Dr Chlebicki.
“That will tell us that this patient may be at risk of developing infection and this patient may need antibiotics,” he added.
DECIDING THE NEED FOR ANTIBIOTICS
Data from about 8,000 patients seen by SGH between 2019 and 2020 was used to build the AI pneumonia model for a pilot study.
This data included X-rays, clinical symptoms, periodic vital signs, and trends of common body responses to infection.
The information was then validated against another 2,000 cases in 2023.
The pilot study found that the system, called Augmented Intelligence in Infectious Diseases (AI2D), accurately determined if antibiotics were necessary upon a patient's first diagnosis in nine out of 10 cases, SGH said.
The use of such technology is expected to reduce over-prescription of antibiotics as part of efforts to fight the growing problem of drug-resistant bugs.
Related:
Dr Chlebicki noted that it is often hard to tell definitively that patients will benefit from antibiotics based on clinical assessment, patient-specific factors, or condition severity alone.
However, dire complications can arise from belated prescriptions of antibiotics for those who genuinely need them.
The current standard is that doctors prescribe antibiotics on suspicion of pneumonia without waiting for confirmation, as some test results may take up to two days, he said.
“We cannot wait that long. If we wait without starting treatment and make a mistake, the patient can possibly get sicker, so the initial day or two in treatment of pneumonia are very crucial,” noted Dr Chlebicki.
“Yet, misusing antibiotics contributes to antibiotic resistance, posing challenges for future infection treatment. A tool like AI2D will be very helpful to guide doctors’ decisions before lab results are available.”
PRESCRIBING ANTIBIOTICS
The cautious attitude towards pneumonia treatment means many patients who were initially suspected to have the infection but later confirmed not to have one, would, in retrospect, have been prescribed antibiotics unnecessarily, Dr Chlebicki noted.
The pilot study revealed that last year, almost 40 per cent of antibiotics prescribed to patients to treat pneumonia at the onset may not have been necessary.
Going forward, the team, which is led by SGH’s Division of Pharmacy, will measure the system’s effectiveness in safely reducing antibiotic use in a comparative study involving 200 inpatients.
These patients will be randomly assigned to doctors who could use AI2D to figure out whether to give them antibiotics.
Related:
TACKLING ANTIMICROBIAL RESISTANCE
Antimicrobial resistance (AMR) - when drugs do not work against viruses, bacteria and fungi - makes infections harder to treat and increases the risks of other medical procedures and treatments, according to the World Health Organization.
These include surgery, caesarean sections and chemotherapy for cancer.
One person is projected to die every three seconds from AMR by 2050 if the global problem is not addressed.
Associate Professor Andrea Kwa, deputy director of pharmacy (research and innovation) at SGH, described AMR as a “silent pandemic” that is “coming up fast and fierce”.
“That led us to think of a solution that will quickly help our doctors decide whether this patient needs antibiotics for his or her pneumonia,” said Assoc Prof Kwa, who led the project.
The process can help doctors save up to 20 minutes per case, she said.
“If we don't have to prescribe antibiotics, we are able to retard the emergence of further resistance,” she added.
While the model undergoes further development, the team hopes it can eventually be used for other common conditions like urinary tract infections, and be expanded to other healthcare institutions.
Continue reading...