User login
TOPLINE:
Artificial intelligence (AI) boosts the screening rate for potentially blinding diabetes eye disorders in a diabetes clinic compared with referral to an eye care provider (ECP) in a racially and ethnically diverse youth population with diabetes.
METHODOLOGY:
- Although early screening and treatment can prevent diabetic eye diseases (DEDs), many people with diabetes in the United States lack access to and knowledge about diabetic eye exams.
- The trial included 164 patients aged 8-21 years (58% female, 35% Black, and 6% Hispanic) with type 1 or 2 diabetes with no known DED and no diabetic eye exam in the last 6 months.
- In a diabetes clinic, patients were randomly assigned to an AI diabetic eye exam (intervention arm) then and there or to standard of care, referred to an ECP with scripted educational material (control).
- Participants in the intervention arm underwent the 5- to 10-minute autonomous AI diabetic eye exam without pharmacologic dilation. The results were generated immediately as either “DED present” or “DED absent.”
- The primary outcome was the completion rate of documented diabetic eye exams within 6 months (“primary gap closure rate”), either by AI or going to the ECP. The secondary outcome was ECP follow-up by intervention participants with DED (intervention) and all control patients.
TAKEAWAY:
- Within 6 months, all the participants (100%) in the intervention arm completed their diabetic eye exam, a primary care gap closure rate of 100% (95% CI, 96%-100%).
- The rate of primary care gap closure was significantly higher in the intervention vs control arm (100% vs 22%; P < .001).
- In the intervention arm, 64% of patients with DED followed up with an eye care provider within 6 months compared with a mere 22% participants in the control arm (P < .001).
- Participants reported high levels of satisfaction with autonomous AI, with 92.5% expressing satisfaction with the exam’s duration and 96% expressing satisfaction with the whole experience.
IN PRACTICE:
“Autonomous AI increases diabetic eye exam completion rates and closes this care gap in a racially and ethnically diverse population of youth with diabetes, compared to standard of care,” the authors wrote.
SOURCE:
This study, which was led by Risa M. Wolf, MD, department of pediatrics, division of endocrinology, Johns Hopkins School of Medicine, Baltimore, was published online on January 11, 2024, in Nature Communications.
LIMITATIONS:
This study used autonomous AI in the youth although it’s not approved by the US Food and Drug Administration for use in individuals aged 21 years and younger. Some of the participants in this study were already familiar with autonomous AI diabetic eye exams, which might have contributed to their willingness to participate in the current study. The autonomous AI used in the study was shown to have a lack of racial and ethnic bias, but any AI bias caused by differences in retinal pigment has potential to increase rather than decrease health disparities.
DISCLOSURES:
The clinical trial was supported by the National Eye Institute of the National Institutes of Health and the Diabetes Research Connection. Wolf, the lead author, declared receiving research support from Boehringer Ingelheim and Novo Nordisk outside the submitted work. Coauthor Michael D. Abramoff, MD, declared serving in various roles such as investor, director, and consultant for Digital Diagnostics Inc., as well as other ties with many sources.
A version of this article appeared on Medscape.com.
TOPLINE:
Artificial intelligence (AI) boosts the screening rate for potentially blinding diabetes eye disorders in a diabetes clinic compared with referral to an eye care provider (ECP) in a racially and ethnically diverse youth population with diabetes.
METHODOLOGY:
- Although early screening and treatment can prevent diabetic eye diseases (DEDs), many people with diabetes in the United States lack access to and knowledge about diabetic eye exams.
- The trial included 164 patients aged 8-21 years (58% female, 35% Black, and 6% Hispanic) with type 1 or 2 diabetes with no known DED and no diabetic eye exam in the last 6 months.
- In a diabetes clinic, patients were randomly assigned to an AI diabetic eye exam (intervention arm) then and there or to standard of care, referred to an ECP with scripted educational material (control).
- Participants in the intervention arm underwent the 5- to 10-minute autonomous AI diabetic eye exam without pharmacologic dilation. The results were generated immediately as either “DED present” or “DED absent.”
- The primary outcome was the completion rate of documented diabetic eye exams within 6 months (“primary gap closure rate”), either by AI or going to the ECP. The secondary outcome was ECP follow-up by intervention participants with DED (intervention) and all control patients.
TAKEAWAY:
- Within 6 months, all the participants (100%) in the intervention arm completed their diabetic eye exam, a primary care gap closure rate of 100% (95% CI, 96%-100%).
- The rate of primary care gap closure was significantly higher in the intervention vs control arm (100% vs 22%; P < .001).
- In the intervention arm, 64% of patients with DED followed up with an eye care provider within 6 months compared with a mere 22% participants in the control arm (P < .001).
- Participants reported high levels of satisfaction with autonomous AI, with 92.5% expressing satisfaction with the exam’s duration and 96% expressing satisfaction with the whole experience.
IN PRACTICE:
“Autonomous AI increases diabetic eye exam completion rates and closes this care gap in a racially and ethnically diverse population of youth with diabetes, compared to standard of care,” the authors wrote.
SOURCE:
This study, which was led by Risa M. Wolf, MD, department of pediatrics, division of endocrinology, Johns Hopkins School of Medicine, Baltimore, was published online on January 11, 2024, in Nature Communications.
LIMITATIONS:
This study used autonomous AI in the youth although it’s not approved by the US Food and Drug Administration for use in individuals aged 21 years and younger. Some of the participants in this study were already familiar with autonomous AI diabetic eye exams, which might have contributed to their willingness to participate in the current study. The autonomous AI used in the study was shown to have a lack of racial and ethnic bias, but any AI bias caused by differences in retinal pigment has potential to increase rather than decrease health disparities.
DISCLOSURES:
The clinical trial was supported by the National Eye Institute of the National Institutes of Health and the Diabetes Research Connection. Wolf, the lead author, declared receiving research support from Boehringer Ingelheim and Novo Nordisk outside the submitted work. Coauthor Michael D. Abramoff, MD, declared serving in various roles such as investor, director, and consultant for Digital Diagnostics Inc., as well as other ties with many sources.
A version of this article appeared on Medscape.com.
TOPLINE:
Artificial intelligence (AI) boosts the screening rate for potentially blinding diabetes eye disorders in a diabetes clinic compared with referral to an eye care provider (ECP) in a racially and ethnically diverse youth population with diabetes.
METHODOLOGY:
- Although early screening and treatment can prevent diabetic eye diseases (DEDs), many people with diabetes in the United States lack access to and knowledge about diabetic eye exams.
- The trial included 164 patients aged 8-21 years (58% female, 35% Black, and 6% Hispanic) with type 1 or 2 diabetes with no known DED and no diabetic eye exam in the last 6 months.
- In a diabetes clinic, patients were randomly assigned to an AI diabetic eye exam (intervention arm) then and there or to standard of care, referred to an ECP with scripted educational material (control).
- Participants in the intervention arm underwent the 5- to 10-minute autonomous AI diabetic eye exam without pharmacologic dilation. The results were generated immediately as either “DED present” or “DED absent.”
- The primary outcome was the completion rate of documented diabetic eye exams within 6 months (“primary gap closure rate”), either by AI or going to the ECP. The secondary outcome was ECP follow-up by intervention participants with DED (intervention) and all control patients.
TAKEAWAY:
- Within 6 months, all the participants (100%) in the intervention arm completed their diabetic eye exam, a primary care gap closure rate of 100% (95% CI, 96%-100%).
- The rate of primary care gap closure was significantly higher in the intervention vs control arm (100% vs 22%; P < .001).
- In the intervention arm, 64% of patients with DED followed up with an eye care provider within 6 months compared with a mere 22% participants in the control arm (P < .001).
- Participants reported high levels of satisfaction with autonomous AI, with 92.5% expressing satisfaction with the exam’s duration and 96% expressing satisfaction with the whole experience.
IN PRACTICE:
“Autonomous AI increases diabetic eye exam completion rates and closes this care gap in a racially and ethnically diverse population of youth with diabetes, compared to standard of care,” the authors wrote.
SOURCE:
This study, which was led by Risa M. Wolf, MD, department of pediatrics, division of endocrinology, Johns Hopkins School of Medicine, Baltimore, was published online on January 11, 2024, in Nature Communications.
LIMITATIONS:
This study used autonomous AI in the youth although it’s not approved by the US Food and Drug Administration for use in individuals aged 21 years and younger. Some of the participants in this study were already familiar with autonomous AI diabetic eye exams, which might have contributed to their willingness to participate in the current study. The autonomous AI used in the study was shown to have a lack of racial and ethnic bias, but any AI bias caused by differences in retinal pigment has potential to increase rather than decrease health disparities.
DISCLOSURES:
The clinical trial was supported by the National Eye Institute of the National Institutes of Health and the Diabetes Research Connection. Wolf, the lead author, declared receiving research support from Boehringer Ingelheim and Novo Nordisk outside the submitted work. Coauthor Michael D. Abramoff, MD, declared serving in various roles such as investor, director, and consultant for Digital Diagnostics Inc., as well as other ties with many sources.
A version of this article appeared on Medscape.com.