Abstract
YouDiagnose carried out a pilot study that seeks to compare the accuracy of expert human predictions with those made by a predictive model. Specifically, this study will analyse the prediction of diseases, cancer risk and the care need of patients being admitted or referred to the specialist service. The doctor was provided with a set of patient data, e.g., patient ID 9001 to 9050. The pilot consisted of 2 parts 1. Clinical prediction and 2. Model validation. In the first part, the doctor had to read the information in the clinical vignette and had to select one of the predetermined choices for the best clinical prediction (DP-1, DP-2, DP-3). Afterwards, he/she was asked to validate the model’s predictions (MP-1, MP-2, MP-3) and recommendations (R-1, R-2, R-3) for the same case.
The study shows that MP has higher total accuracy (82.8%) compared to DP (50.6%). In predicting cancer, the MP method has higher sensitivity (100.0%) and positive predictive value (38.5%) compared to the DP method (90.0% sensitivity and 30.0% positive predictive value). The MP method also has higher specificity (78.9%) compared to the DP method (72.7%). Both methods have high negative predictive values (98.2% for DP and 100.0% for MP) with a p-value of 0.3705.
The results of the current pilot study demonstrate the model’s potential, while also highlighting areas where further testing is needed in order to increase user confidence and improve the accuracy of diagnosis. Such testing could provide invaluable insights into how to maximize the value of the system in offering better frontline screening solutions e.g., triaging, clinical decision support, risk-based clinic booking system etc.
Competing Interest Statement
The authors have declared no competing interest.
Funding Statement
This study was funded by YouDiagnose Limited
Author Declarations
I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.
Yes
The details of the IRB/oversight body that provided approval or exemption for the research described are given below:
At the outset, the research methodology was approved by YouDiagnose Ethical Approval Committee
I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.
Yes
I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).
Yes
I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.
Yes
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Glossary of terms
- 1. Similarity match
- ‘Similarity match’ term is used when there was a match with a disease which is similar in etiopathogenesis, sharing the presenting features and at least the initial line of treatment. For example, certain types of diseases cannot be distinguished from one another based on presentation, and it is almost impossible to distinguish between them based on medical history alone. Needs scrutiny. However, there are many diseases that share similar aetiology, pathogenesis, symptoms, initial plan, and treatment. Examples can be given of lactation mastitis and breast abscess which are infections involving breast tissue in a breastfeeding woman and need prompt treatment with antibiotics. Both have the same symptoms and cannot be distinguished without ultrasound.
- 2. Gold standard
- For all cases, the final diagnosis or final diagnosis was accepted as the disease the patient had, as confirmed from the patient’s case record after the completion of investigations and interventions. This was taken as the gold standard for comparing model performance.
