Views on artificial intelligence (AI) assisted clinical trials

It is of interest to document the views of medical professionals on the application of artificial intelligence (using known data for the prediction of unknown events) in clinical trials using a web survery with a structured questionnaire from 377 subjects. The questionnaire contained 17 statements which were categorised into awareness (1,2 statements), perception (3-10 statements) and opinion (11-17 statements). The data obtained was compared between the subjects using two tailed Fisher's exact test with p-value <0.05 for data significance analysis. Data shows that majority of professionals have possitive views on the application of artificial intelligence in clinical trials. This will accelarrate the drug evaluation process. However, the use of emerging tools such as AI will not replace human subjects in this context.


Background
New drugs require more than 10 years to reach the market [1]. Hence, investment in drug design, research, development and formulation comes with high risk for pharma companies [2,3]. Therefore, the use of artificial intelligence help clinicians and researcher to identify specific targets in this context [4].

Materials and Methods:
This study was conducted by BioHymns Innovations Pte. Ltd, Singapore, during Dec 2019 to May 2020 in Tamil Nadu, India. The sample size was measured using the Raosoft online calculator (Raosoft) and included 377 subjects. Medical doctors, who are involved or have been involved in clinical trials as investigator or co-investigator and medical Doctors who are interested in participating in this online survey were included in the study. Medical doctors who are not involved or have not been involved in clinical trials, non-medical doctors and medical doctors not interested in participating this online survey were excluded from the study.

Study procedure:
This was a questionnaire-based study. Regarding questionnaire validity and reliability, a structured questionnaire was developed after a thorough literature review, which was conducted initially by the chief investigator and research papers were shortlisted for further discussion among the research team. All the views, thoughts and concerns on the proposed study were taken into consideration during the design phase. An initial draft of the questionnaire was designed after the research team had reviewed all the selected papers comprehensively. Individual survey items were reviewed by a group of medical professionals and consensus were reached regarding the clarity and importance of each item. The validation process was further expanded by piloting the questionnaire with four experienced doctors who meet the eligibility criteria and are not aware of this study. There was voluntary participation by the physicians. The questionnaire was framed in English. The questionnaire comprised 17 statements which were sectioned in to 3 events like awareness (1,2 statements), perception (3-10 statements) and opinion (11-17 statements). After obtaining ethics committee approval VISTAS-SPS/IEC/VIII/2019/04, the questionnaire was shared to professionals involved in clinical trials using survey Google forms.

Statistical analysis:
An anonymous questionnaire was shared through Google forms to all participants. Basic statistics for the responses was done and represented as total number and percent. The data obtained was compared between the specialities using two tailed Fisher's exact test. P-value <0.05 was taken as significant.

Results:
This study included 377 participants comprising resident doctors (N= 143), doctors working as clinical research associates (N= 12), paediatricians (N= 7), general physicians (N= 47), pharmacologists (N= 161), and clinical trial physicians (N= 7). The questionnaire consisting 17 statements have been categorized in to 3 types as for awareness, perception and opinion. The statements in the questionnaire were enlisted in Table 1. The response rate towards the questionnaire statements was 100%.

Responses against questionnaire statements
The consolidated responses were tabulated in Table 1. To describe the responses in general, majority (83.5 % & 65.5%) of the participants were aware of the AI based health care delivery and clinical trials. Most of the participants identified the potentiality of AI in, clinical trial processes, time saving or accelerating drug development, cost-effectiveness, and handling vast data. The AI based clinical trials was supported by large number of participants, but some has suggested that AI cannot substitute human intelligence and also, might raise ethical and legal concerns.

Awareness
Statements 1 and 2 were categorised for analysis of awareness in this study. The responses were analysed as per the category of speciality. 220 positive responses out of 286 towards awareness were obtained from resident doctors. 16/24, 9/14, 66/94, 242/322, and 9/14 positive responses were obtained from clinical research associates, paediatricians, general physicians, pharmacologists, and clinical trial physicians respectively (Figure 1).