Hi,
We have few AI (Artificial Intelligence) solutions for different problems in public health. Few of the problems are binary in nature, while the rest is continuous. We need help in calculating sample size for measuring the accuracy of the AI (to reliably predict the problem).
For example, we developed an AI solution to estimate weight of a baby. We expect the AI to predict the weight reliably in 90% of babies - error to be less than 10% of actual weight by gold standard equipment. I can calculate sample size in two ways, I think:
What should we choose? In the second option, which SD should we choose for ss calculation?
Thanks for reading and suggesting in advance.
PS - both the methods are applied on same study participant.
Dear sir,
Download the Aidos-x system for free from the site http://lc.kubagro.ru/aidos/_Aidos-X.htm . It is an excellent intelligent system for diagnosing and classifying human diseases.
Methods for using the system Aidos-x for diagnosing human diseases are disclosed in lectures with sound "Using automated system-cognitive analysis for the classification of human organ tumors", "Intelligent system for diagnosing early stages of chronic kidney disease", which can be downloaded right now from the website https://www.patreon.com/user?u=87599532 After subscribing to this site, you will receive databases for medical research to identify those diseases that you will read about in lectures. The acquired skills in working in the system Aidos-x will allow you to apply for grants to perform scientific research in medicine.
Sincerely, Vladimir Ryabtsev, Doctor of Technical Sciences, Professor of Information Technology
To evaluate AI solutions in health, one should consider the following factors:
Performance: Evaluate the accuracy, precision, recall, and F1 score of the AI solution in comparison to traditional methods or benchmarks.
Data quality: Assess the quality of data used to train and test the AI model, including data volume, diversity, and representativeness.
Clinical validation: Verify the efficacy and safety of the AI solution in clinical trials or studies conducted with patients.
Interpretability and explainability: Ensure the AI solution is transparent and explainable, and that it can provide clear explanations of its decision-making process.
Deployment and integration: Evaluate the feasibility of integrating the AI solution into clinical workflows, and assess the costs and benefits of implementing the solution.
Ethical and legal considerations: Address ethical and legal implications of the AI solution, such as privacy, confidentiality, bias, and liability.
Dear Mr Shah
If I were you, I would use the free tool Gpower (https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower). You can add expected effects as well as the type of the tests that you are going to run. The tool will calculate for you the estimates sample size.
There are also rules of thumb for SEM and similar methods, as well as for randomised experiments. But with this tool, you should be able to calculate the sample sizes correctly.
Good luck with your research!
Sicnerely
Dr Yevgen Bogodistov
The principles of using automated system-cognitive analysis for the diagnosis of human diseases are disclosed in presentations with sound in English, which are posted on the site https://www.patreon.com/user?u=87599532. By subscribing to this site, you will receive free of charge representative databases for diagnosing heart disease, early stage kidney disease and detecting tumors in human organs.
- Badges
- Science topic