Study finds AI tool MySTIRisk effectively identifies high-risk STI subgroups


A recent study revealed the effectiveness of the AI tool MySTIRisk in pinpointing high-risk subgroups for HIV and other sexually transmitted infections, shedding light on improved risk assessment and testing accessibility.

Study finds AI tool MySTIRisk effectively identifies high-risk STI subgroups | Image Credit: © Quardia Inc. - © Quardia Inc. -

Study finds AI tool MySTIRisk effectively identifies high-risk STI subgroups | Image Credit: © Quardia Inc. - © Quardia Inc. -

The artificial intelligence (AI) tool MySTIRisk is effective at identifying subgroups at high risk of HIV or other sexually transmitted infections (STIs), according to a recent study published in Open Forum Infectious Diseases.


  1. The AI tool MySTIRisk has demonstrated effectiveness in identifying high-risk subgroups for HIV and other sexually transmitted infections (STIs), offering potential for targeted interventions and preventive measures.
  2. With approximately 376 million new cases of curable STIs annually, there's a pressing need for effective risk assessment tools due to adverse health outcomes and substantial economic costs associated with these infections.
  3. MySTIRisk represents an advancement in risk prediction by utilizing machine learning algorithms, offering concurrent assessments and potentially more accurate predictions compared to traditional logistic regression models.
  4. Determining optimal cutoff points is crucial for providing appropriate recommendations. The study used Youden's index to establish cutoff points for MySTIRisk risk scores, enabling the categorization of consultations into binary risk groups.
  5. High-risk subgroups, particularly men who have sex with men, showed elevated positivity rates for STIs. This underscores the importance of targeted interventions and tailored testing strategies for vulnerable populations identified by MySTIRisk.

Approximately 376 million new cases of curable STIs are reported among sexually active individuals per year, leading to adverse health outcomes and significant economic costs. A lack of knowledge about personal STI risk, lack of testing accessibility, and social stigma prevent many patients from seeking testing and treatment.

While risk protection tools are available to help patients determine their personal STI risk, many of these tools use logistic regression methodology instead of providing concurrent assessments using more advanced algorithms. A machine learning tool called MySTIRisk has been developed to improve prediction by enhancing model performance for STIs.

To provide appropriate recommendations, optimal cutoff points must be established. Investigators conducted a retrospective cross-sectional study to determine optimal cutoff points for MySTIRisk risk scores.

Participants included patients with a confirmed diagnosis from the Melbourne Sexual Health Centre between January 2008 and May 2022. Self-reported demographic and sexual behavioral data was obtained from computer-assisted self-interviewing at baseline and follow-up visits, with at least 3 months between follow-ups.

Additional data was extracted from electronic records, and included 216,252 HIV consultations, 227,995 syphilis consultations, 262,599 gonorrhea consultations, and 320,355 chlamydia consultations. Machine learning models were trained on demographic, behavioral, and diagnostic data.

Risk scores ranged from 0 to 1, with 1 indicating the greatest risk. Age, gender, country of birth, men reporting having sex with other men, condom use, number of partners, STI symptom presence, past STIs, injection drug use, contact with STI diagnoses, and sexual partners outside Australia or New Zealand were reported as key predictors.

Youden's index, a common metric for identifying optimal cutoff points in risk models, was used to determine the optimal cutoff point for the MySTIRisk model. Possible cutoff values were estimated at intervals of 0.02. Once an optimal cutoff point was determined, consultations were categorized into binary risk groups.

Men who have sex with men (MSM) accounted for 40% to 50% of consultations, women for 30% to 36%, and heterosexual men for 15% to 25%. Participants were aged a median 29 years, and almost half were born overseas while the remaining were from Australia and New Zealand.

The HIV data set had a median risk score of 0.32, the syphilis data set 0.35, the gonorrhea data set 0.37, and the chlamydia data set 0.42. The optimal cutoff points for these infections were 0.56, 0.49, 0.52, and 0.47, respectively, with sensitivity and specificity values of 86% and 65.6%, 77.6% and 78.1%, 78.3% and 71.9%, and 68.8% and 63.7%, respectively.

High-risk was found in 35% of HIV consultations, 23% of syphilis consultation, 31% of gonorrhea consultations, and 39% of chlamydia consultations. The high-risk group accounted for 86%, 78%, 78%, and 69% of infections, respectively.

Overall positivity was 8.1% for chlamydia, 5.9% for gonorrhea, 1.7% for syphilis, and 0.3% for HIV. Among high-risk groups for the infections, the positivity rates were increased by 3.9, 9.2, 12.3, and 11.7, respectively, when compared to the average-risk groups.

MSM had the highest positivity while women had the lowest. Positivity was impacted by differing testing recommendations between groups.

These results indicated efficacy from the MySTIRisk tool in determining high-risk subgroups for STIs such as HIV. Investigators recommended further research about expenses associated with HIV testing.


Latt PM, Soe NM, Xu X, et al. Identifying individuals at high risk for HIV and sexually transmitted infections with an artificial intelligence–based risk assessment tool. Open Forum Infectious Diseases. 2024;11(3). doi:10.1093/ofid/ofae011

Related Videos
Why doxycycline PEP lacks clinical data for STI prevention in women
Understanding the impact of STIs on young adults | Image Credit:
Raising awareness on STDs
Related Content
© 2024 MJH Life Sciences

All rights reserved.