Artificial intelligence in outcomes research: a systematic scoping review

ABSTRACT

Introduction: Despite the number of systematic reviews of how artificial intelligence is being used in different areas of medicine, there is no study on the scope of artificial intelligence methods used in outcomes research, the cornerstone of health technology assessment (HTA). This systematic scoping review aims to systematically capture the scope of artificial intelligence methods used in outcomes research to enhance decision-makers’ knowledge and broaden perspectives for health technology assessment and adoption.

Areas covered: The review identified 370 studies, consisted of artificial intelligence methods applied to adult patients who underwent any health/medical intervention and reported therapeutic, preventive, or prognostic outcomes. Artificial intelligence was mainly used for the prediction/prognosis of more frequently reported outcomes, efficacy/effectiveness, among morbidity outcomes. The predictive analysis was common in neoplastic disorders. Neural networks algorithm was predominantly found in surgical method studies, but a mixture of artificial intelligence algorithms was applied to the studies with the rest of the interventions.

Expert opinion: There are certain gaps in artificial intelligence applications used in outcomes research across therapeutic areas and further considerations are needed by decision-makers before incorporating artificial intelligence usage into HTA decision-making processes.

KEYWORDS:

• Artificial intelligence
• decision-making
• machine learning
• outcomes research
• systematic review

Acknowledgments

We would like to thank Dr. Nazila Assasi for her contribution to the scientific content of this article and also resolving the conflict between the reviewers, Dr. Farhang Modaresi for his contribution to screening and comments on the manuscript, Mr. Bijan Farhoudi for his contribution to identify the primary purpose of using AI, Ms. Maria Eberg for her comments related to the grouping of AI algorithms, Dr. Hoda Moin Sadat for her collaboration on grouping the morbidity outcomes and Dr. Jason Brownlee for his permission to reprint his machine learning framework.

Author Contributions

• Conception and design – Mainly Pooyeh Graili and Luciano Ieraci

• Screening and reviewing the included articles and data extraction – Mainly Pooyeh Graili, Luciano Ieraci and Nazanin Hosseinkhah

• Analysis of the data – Mainly Pooyeh Graili, Luciano Ieraci and Mary Argent-Katwala

• Interpretation of the data – All authors

• Drafting of the paper or revising it critically for intellectual content – All authors

• Final approval of the version to be published – All authors

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

The List of Abbreviations

Table

Reviewers Disclosure

Peer reviewers on this manuscript have no relevant financial relationships or otherwise to disclose.

Prior Abstract Presentations

The content of this work has been presented in the format of two abstracts at SMDM congress and CADTH symposium. The abstract submitted to CADTH had also been accepted as poster presentation in the EU ISPOR Congress, but we have withdrawn because it would be unethical to present one abstract in two events.