|Dr. Philip C Anderson in Dermatology Department circa 1967 (Photo credit: Wikipedia)|
We are again having an interesting discussion on new generation hairloss treatments in the Dermatologists Sans Borders facebook group. We have discussed about the inconsistencies in the claims of the so called biomimetic peptides before.(Beauty born out of Bioinformatics) So we will not go back into that again. One of the senior members pointed out the two fundamental limitations in conducting a proper clinical trial in a non-critical intervention like hairloss remedies. Cost and time. So this post is basically to explore the challenges and the probable solution in this scenario. I am not a subject expert and the intention is just to offer few breadcrumbs from other fields to ignite thinking amongst this group.
Historical control data:
Historical control data consists of a wide range of data gathered from diverse experiments performed under widely differing conditions. Though historical control data is not as reliable as concurrent controls, it may suffice in this situation as the intervention is non-critical. Use of historical control data may significantly reduce the overall trial costs.
Single-subject design is a research design in which the subject serves as his/her own control, and is sensitive to individual organism differences important in cosmetic interventions. It may also reduce the interpretation bias and reduce the time required to arrive at a preliminary conclusion.
Equivalence / Non-inferiority trials:
Equivalence trials are designed to demonstrate that one treatment is as effective as another. This is especially important here as exotic and expensive treatments bust into the scene all the time. In my cynical view, most of these interventions may not even clear a non-inferiority trial
Systematically reviewing and synthesising evidence from conversation analysis.
This is a new concept that is emerging and may become more and more significant in the new digital age of social networking. To quote from an article in BMC Medical Research Methodology:
Healthcare delivery is largely accomplished in and through conversations between people, and healthcare quality and effectiveness depend enormously upon the communication practices employed within these conversations. An important body of evidence about these practices has been generated by conversation analysis and related discourse analytic approaches, but there has been very little systematic reviewing of this evidence.
The Dermatologists sans Borders facebook group was used as a cohort for research couple of years back. Hope this strictly curated group of dermatologists contribute to the corpora of knowledge by active discussions like this.
1. Haseman, Joseph K. “Data analysis: Statistical analysis and use of historical control data.” Regulatory Toxicology and Pharmacology 21.1 (1995): 52-59.
2. D’Agostino, Ralph B, Joseph M Massaro, and Lisa M Sullivan. “Non‐inferiority trials: design concepts and issues–the encounters of academic consultants in statistics.” Statistics in medicine 22.2 (2003): 169-186.
- Named Entity Recognition using LLMs: a cTakes alternative? - September 1, 2023
- Distilling LLMs to small task-specific models - August 24, 2023
- Kedro for multimodal machine learning in healthcare - January 25, 2023