Karibu tena! The last time you tuned in, we were headed to Arusha to begin a four-day, three-night safari across three of Tanzania’s national parks. We’ve now been back for over a week, and I mourn the experience every single day.
On the first day, we visited Tarangire National Park, which I personally think was the best day of the safari. We saw countless elephants, as the park is known for its elephant sightings, but one unexpected highlight was a baboon troop moving along the road we were traveling on. We estimated around 250 baboons, and the best part was watching the babies ride on their mothers’ backs like little jockeys.



And then, of course, there were the countless lions, including cubs. I didn’t get a good picture of the cubs because I was too busy reacting to every tiny movement they made. Lions are incredibly lazy, just like house cats, so most of what we saw was napping, but I was enthralled nonetheless.
We had two very different nights in the Serengeti. The first was at a luxury tented safari camp, still not sure how that fit in our budget, and it was such a cool experience. The dinner tent was a short walk from our sleeping tents, but after dark, staff insisted on escorting us, just in case we ran into lions or hyenas. That made it feel so otherworldly. I was genuinely in awe, and a little in disbelief, that this was real life.

An elephant also wandered into our campsite, though maybe we were the ones who wandered into his. It’s crazy I can say I stood so close to an elephant without any barriers.
On the last day, we visited the Ngorongoro Crater and saw even more wildlife. A few highlights were the lone hyena strolling across the landscape, the warthogs, and especially the wildebeests, which honestly look so fake to me. As our safari wrapped up, I had a total epiphany moment, standing on my seat with the roof popped open as the safari truck sped through the plains, thinking about how incredibly lucky I am to have witnessed this kind of nature. Forever grateful.



As for my project, the week before last was full of important decisions. As I mentioned in my previous blog, there are a few limitations in the ambient temperature readings collected by the Hadli monitors. After investigating further, we found that most monitors are installed near the ceiling and that each unit typically has only one monitor, even in facilities with multiple rooms. In some cases, this means the readings may not fully represent the temperatures experienced by the majority of newborns.
That said, we do not discredit the data at all. These measurements are incredibly valuable in providing an initial picture of ambient conditions. Our goal is simply to build on this foundation to get the most accurate possible estimate of what the broader newborn population is likely experiencing. To do this, we decided to adjust the recorded data by 2–4°C, since the original numbers likely reflect air higher up in the room. My mentor John helped make this adjustment by simulating how temperature typically changes from floor to ceiling. Because colder air tends to settle near the floor, especially by doors or windows, he modeled this drop and factored in cot height, subtle room variations, and small sensor inaccuracies to generate an approximation of what babies might feel.
The graphs below compare monthly trends in facility-level minimum daily temperatures before and after adjusting the Hadli data. Each graph categorizes facilities based on whether their daily minimum temperatures remained within, exceeded, or fell below the ideal 25–35°C range. Each bar is color-coded: green indicates that a facility maintained temperatures within range every day of that month, yellow indicates at least one day above the range, and blue indicates at least one day below it. In the original Hadli data, green dominates the chart, suggesting that most facilities were able to maintain temperatures within the ideal range. However, after the adjustment, the proportion of green decreases and blue increases, revealing that more facilities may actually be experiencing days with below-range temperatures than previously observed.
The country name has been omitted from the chart.

We also revisited a previous decision regarding how we account for ambient temperatures after admission. Initially, we used an average from the seven days post-admission, but we decided to shift to using the median length of stay per country (Malawi, Kenya, Nigeria, and Tanzania). This led to slight improvements in our model.
Around this time, I began preparing for my final presentation with the KIT Key Interventions Technical Working Group at NEST. This group facilitates cross-country collaboration to spark discussion and catalyze programmatic action. I was nervous but also excited about the opportunity to share my findings.
The timing worked out perfectly, as my mentor Christine was traveling to Nairobi, Kenya to meet with the NEST team there, and she graciously invited me along. I’m so glad she did. Nairobi stole my heart. The city has so much to offer, and the Kenya team was incredibly warm and welcoming. Every day I had the chance to meet inspiring people who were contributing their unique talents to the same mission. I received incredibly valuable feedback ahead of my presentation, and felt truly supported.
John, my mentor and NEST data analyst, had especially helpful feedback on how to present statistical findings to a general audience. He was so patient with my many questions. Hannah also offered encouragement and great insight. Being surrounded by people who genuinely care about their work, collaboration, and my growth has been so meaningful. I know I’ll seek mentors and work environments like this for the rest of my life.

I’m sad that my in-person time is coming to an end, but so excited that my work with NEST will continue even after I return to Houston. I’m so thankful to Christine for believing in me and supporting my continued involvement. Nairobi gave me both a personal and professional experience I’ll never forget.
My presentation was yesterday, and it went really well thanks to my amazing team. I wanted to summarize a few of the key takeaways and the feedback I received. We built a model to assess associations between adjusted ambient temperature and newborn hypothermia after admission. We found a statistically significant association, both before and after temperature adjustment.
Temperatures above 28°C had a protective effect against hypothermia (OR = 0.902; 9.8% reduction in odds compared to 27–28°C). Temperatures at or below 26°C were associated with increased odds of hypothermia (OR = 1.077; 7.7% increase compared to 27–28°C). These tertiles (≤ 26°C, 27–28°C, > 28°C) and were created to ensure groups had roughly equal sample sizes. That said, the effect sizes are small, and we must be cautious about making recommendations. While warmer temperatures may be beneficial, the results should be interpreted with care.
We also confirmed that ambient temperature alone doesn’t explain the hypothermia burden. Other variables like birth weight, birth location, month of admission, and heater availability help explain additional variability. Our takeaway is that clinical practices, especially maintaining the warm chain, are essential. But more covariates and further data, like satellite temperature trends, may improve the model even further.

Overall, the presentation was a very positive and thought-provoking experience. I’m excited to incorporate the feedback and explore new directions for the project. This week, I’ll also begin drafting the methods section for our eventual publication, and I’m looking forward to summarizing my work in detail and reflecting on this journey professionally.
This is actually my final entry from Dar es Salaam, and I hope to make the most of my last week here. Thank you for reading my blog this summer. Tuonane!