Let me tell you a short story...
 |
A red-stained schistosoma worm. The male is the small one,
spending his entire life within the groove of the female. P/C CDC. |
As you know, I’m writing another book–this one is about Unanticipated Consequences of the choices we made. (Here’s my substack that I’m using to try out various sections.)
And, as you also know, I have a deep interest in Egypt. So, on a recent visit there, I visited the Aswan High Dam (which we discussed earlier in SRS in conjunction with its creation of Lake Nasser) and thought about what were some of the Unanticipated Consequences of building that massive dam? This led to our SRS Research Challenge about “What has been the effect of the creation of Lake Nasser on the incidence of schistosomiasis in Egypt?”
A little background: Schistosomiasis (aka snail fever, bilharziasis) is a disease caused by several species of Schistosoma flatworms, any of Schistosoma haematobium, S. japonicum, and S. mansoni. Transmission can occur when humans are exposed to freshwater sources that are contaminated with Schistosoma parasites.
People infected with Schistosoma flatworms shed eggs in their urine or stool. In fresh water, immature flatworms called miracidia hatch from eggs. The miracidia find and infect the tissues of freshwater snails where they mature into an infective stage known as cercariae. The free-swimming cercariae are released from the infected snail into the surrounding water. Cercariae enter their human host by burrowing though skin that is exposed to contaminated water. Male and female parasites migrate to either the liver, the lower intestines, or bladder, where they release eggs into feces or urine.
Symptoms of schistosomiasis are a result of the body’s immune response to the flatworms and their eggs. Symptoms include itching at the cercariae entry sites, fever, cough, abdominal pain, diarrhea, and enlargement of the liver and spleen. In some cases, infection may lead to lesions in the central nervous system.
In that previous post we did a little comparison of the different Deep Research tools out there. At least, we looked at how Deep Research search tools might work. I promised to look at some other analysis tools. This is my briefing on using Google’s NotebookLM.
Full disclosure: Some of my Googler friends currently work on NotebookLM. I've tried to not let that influence my writing here.
How NotebookLM works: The core idea of NotebookLM (NLM) is that you create a notebook on a given topic, download a bunch of sources into it, and then “have a conversation” with your sources.
Think of it as though you’ve got a research assistant–you give them a pile of content (Google Docs, Slides, PDFs, raw text notes, web pages, YouTube videos, and audio files). Then you can chat with them about the stuff that’s in there. It’s as though NLM has just read everything and understands it reasonably well. This is handy if you’ve got a lot of content to use. (In the AI biz this is called “Retrieval Augmented Generation,” which means that the text generation process is driven primarily by the contents of the sources you’ve uploaded.)
So… by using our standard SRS methods, I searched for good online content (mostly PDFs of medical articles, mostly found by using Google Scholar) and uploaded it to NLM.
I’ll spare you the long story here… but I spent WAY too much time finding low quality content that initially looked good, but turned out to be puff pieces published in low quality journals. (I assume that you, our SRS regulars, know how to find this out!)
I ended up with a screen that looked like this (the red rounded-rectangles and circuled numbers are mine):
 |
| Click to enlarge for details |
In panel 1, I’ve uploaded 9 different papers–one of which was my SRS post from last month about Egypt. I picked the 8 papers as being high quality about the effects of the building of the Aswan Dam on the incidence of Schistosomiasis in Egypt.
Once I loaded them up, I chatted with NLM about the papers a bit in panel 2 and then asked my key question:
[ did the incidence of schistosomiasis increase or
decrease after the Aswan Dam was built? ]
And then:
[ What were the predicted effects of the Aswan High
Dam on schistosomiasis? ]
NLM then gave me a quite reasonable answer based on the documents I’d uploaded as sources. As you can see in the image below, NLM gives citations to places where it extracted the text when you roll over the gray citation number.
 |
Rolling over one of the citation number causes a pop-up that shows exactly where the
supporting text is in the list of sources. |
Now, I have to admit that I read these 9 papers, so I have a good idea what’s in them. And I’d written a draft of the answer to that question. Somewhat to my surprise, the answer was quite good–it’s very much what I would have written (although perhaps not in the style that you see. Here’s the start of the NLM reply to my question (and a link to the full doc):
After the Aswan High Dam (AHD) was built, the incidence of schistosomiasis in Egypt decreased overall, despite initial concerns that the dam would cause it to increase [1-4].
Initial Concerns: It was predicted that the impoundment of the Nile would create ideal conditions for snails, which are intermediate hosts for the schistosomiasis parasite [5-7]. It was also thought that steady water flow downstream from the dam would increase human contact with the water, leading to more contamination [5, 6].
Public Health Measures: The implementation of public health programs played a critical role in reducing schistosomiasis [4, 8-10]. These programs included:
Installation of clean fresh water pumps [11, 12].
Treatment of canals and drainage ditches with copper sulfate [11, 12].
Mass distribution of the drugs metrifonate and praziquantel [11-14].
Mollusciciding to reduce snail populations [15].
… etc …
This is pretty good, but I have a couple of quibbles.
The citations are both wonderful and annoying. The wonderful part comes from being a very specific citation. As you see in the above image, NLM tells you the paper it found AND specifically the supporting text. That’s really nice.
On the other hand, you have to go through the note citation-by-citation, one at a time, to figure out that 1, 3, 5, 6, and 8 all are to the paper “Effects of the Aswan Dam.pdf” in the sources. And, at this time, I haven’t been able to figure out how to get NLM to give me a full citation list–you know, something that looks like this:
Abd-El Monsef, H., Smith, S. E., & Darwish, K. (2015). Impacts of the Aswan high dam after 50 years. Water Resources Management, 29, 1873-1885.
I hope that gets added to the list of things NLM will do, but at the moment, it’s kind of a laborious process to figure out the mapping from citation numbers to actual papers.
But perhaps the most impressive piece of this analysis by NLM was the summary (emphasis by NLM):
In summary, while the Aswan High Dam did cause a change in the types of schistosomiasis seen in Egypt, and there were initial fears the disease would increase, overall schistosomiasis decreased significantly because of public health interventions [9]
That’s the conclusion I came to as well, and it’s definitely NOT the first thing you’d find with just a simple Google search. In fact, before the dam was built, an increase in schistosomiasis was predicted by epidemiologist Van Der Schalie who wrote: “...there is evidence that the high incidence of the human blood fluke schistosomiasis in the area may well cancel out the benefits the construction of the Aswan High Dam may yield (Farvar and Milton 1972)”. It was widely thought that schistosomiasis would increase as farmers converted from basin irrigation system to perennial irrigation and so had more water to irrigate with.
But with such predictions widely known, the Egyptian government began a far-ranging program of killing snails, cleaning up waterways, and giving much of the population chemotherapy against schistosomiasis.
In 1978, soon after the AHD was commissioned, a baseline epidemiological survey on over 15,000 rural Egyptians from three geographical regions of Egypt (Nile Delta, Middle Egypt and Upper Egypt), plus the resettled Nubian population showed that the prevalence of schistosomiasis was 42 % in the North Central Delta region, 27 % in middle Egypt and 25 % in Upper Egypt. That sounds bad, but it’s a massive reduction in disease rates.
So, overall, the control program worked pretty well, with infection rates dropping to record lows.
The thing is, the predictions about the consequences about building the dam were right–but they counteracted the increase by anticipating the consequences and being proactive about fixing them.
Wait! There’s more:
In addition to this chat / query process, you can also ask NLM to create a study guide, an overall summary, or a briefing doc. NLM gives you prompts to use (roughly “Create a study guide from these sources…” or “Summarize these sources in a readable format.”)
As an expert SearchResearcher, you can also chat with just one document to get JUST that document’s perspective (and, implicitly, that author’s point-of–view).
Give this a try!
SearchResearch NotebookLM tips:
Make sure the content you upload as sources is high quality. If you have low quality content, that will surface in the answers you ask NLM.
If you lose the connection between the source you uploaded to NLM, it’s a pain to restore. You have to remember it. Be sure the uploaded source has the identifying info in it. (I always put the citation and the original source URL at the top of the source text.)
If you ask a question about something that’s NOT covered by the sources, the quality will drop. Stay on topic and try to not overdrive your headlights.
I haven't tried using NLM without carefully selecting the sources that go into the collection, but I can imagine a use-case where you add multiple sources in and then ask for the plus-and-minus analysis. That's an interesting experiment--let us know if you try it.
Keep searching.
