The paper reports an observational 12-month study in children and young adults with type 1 diabetes who self-initiated a do-it-yourself AID system shortly after T1D onset.

While the study contains a considerably large cohort and reports data from resource-limited settings, I believe it lacks originality and contains several methodological weaknesses (that the authors mainly correctly identified) that preclude its publishing in Diabetes, Obesity & Metabolism.

Major comments:

1) The study is a single-arm observational study. This design inadvertently leads to several possibilities of bias. Given that there are more RCTs published with similar conclusions, I do not believe the content is original enough for publication.

2) The authors acknowledge the selection bias but they did not mention a very intensive nature or their telehealth visits (which should, in my opinion, be called an intervention themselves).

With such an intensive education, we can expect anyone to improve their results making in impossible to assess whether it was indeed the AID system that was the only reason for improvement.

3) The total length of follow-up is very short to be able to make strong claims regarding the durability of the findings. Indeed, some participants were lost to follow-up while others did not reach the 12-month timepoint further emphasizing this.

I thank the authors for the review which acknowledged some of the objections I have raised - the authors expanded the limitations part of the Discussion to better explain the biases likely present in their study.

Yet, despite this effort, I still think that the methodology (single-arm study with clear selection bias and added telehealth intervention) is weak and the findings are not novel enough to balance this. Therefore, I do not recommend this study for publication in DOM.