At Bewgle, we take immense pride in our NLP capabilities on any unstructured data. Though we have primarily focused on drawing insights from feedback or similar text, I, as a data enthusiast, wanted to challenge the system beyond feedback. One of the use cases that I wanted to try was that of deriving insights from the blogs fetched through Google search results.
“Lowering cholesterol naturally” has been a topic of personal interest to me in the last 6 months and based on my research I have seen positive affirmative results. I wanted to see if I could deduce similar conclusions from Bewgle insights.
I ran a Google search on “how to lower cholesterol naturally” and dumped the data from the first 10 results into Bewgle. As part of our analysis, we extract topics and their sentiments from text. My first thought was that in this particular use case, ‘sentiment’ would not make sense, so I considered ignoring that aspect of the analysis. However, I was in for a surprise with the results.
For the purpose of readability, I have limited the topics under each section to five. Evidently, the topics listed under ‘very positive’ help in lowering cholesterol whereas the ones under ‘very negative’ must be avoided since they have adverse effects on the cholesterol levels.
This was a good snapshot and an easy summary of what you must increase or decrease from your diet. Interestingly, it is synchronous with my personal research findings as well.
To further understand what comprises some of these topics like nuts, seeds, legumes, let us see what food items are clustered under these topics:
|Fruits||Apple, Berry, Pear, Strawberry|
|Veggies||Okra, Eggplant, Broccoli|
|Nuts||Walnut, Almond, Tree Nuts|
|Grains||Brown Rice, Barley, Oat, Quinoa|
|Legumes||Kidney Beans, Chickpea, Soybean, Pinto Bean, Edamame, Lentils|
|Fish||Mackerel, Herring, Sardine, Tuna, Halibut, Salmon, Trout|
|Seeds||Flax Seeds, Chia Seeds|
|Oils||Palm Oil, Coconut Oil, Olive Oil, Safflower Oil|
Each of these subtopics has a sentiment, which can help us understand at the subtopic level the food items that are beneficial vs harmful for cholesterol levels.
For example, let us look at the subtopics under ‘Oils’:
Looking at the ratings, ‘Safflower Oil’ and ‘Olive Oil’ rate higher compared to ‘Hydrogenated Oils’, ‘Palm Oil’ and ‘Coconut Oil’. We can infer that Safflower and Olive oils are recommended, whereas, it would be better to avoid Hydrogenated, Palm and Coconut oils. According to me, this has been a pretty good discovery. Infact, earlier in my research, due to my bias and conditioning, I seem to have ignored Safflower oil, but the analysis here forced me to revisit that aspect. I didn’t have to look much further in the system to figure out why. Here are a few relevant summaries that the system had provided that were helpful to understand:
- “In fact, regular consumption of Safflower Oil is tied to lower total cholestrol and LDL Cholestrol compared to that of Olive Oil making it a good go-to cooking oil.”
- “Safflower oil – This neutral, high-heat oil is rich in Phytosterols, Cholestrol-Blocking plant compunds that could lower your LDL cholestrol by as much as 14% accordinig to the Clevaland Clinic.”
- “You can saute vegetables in Olive Oil, add it to a marinade or mix it with vinegar as a salad dressing.”
- “You can also use olive oil as a substitute for butter when basting meat or as a dip for bread.”
As can be seen from the snapshot generated by Bewgle above, Safflower oil seems to have a high-heat point compared to Olive oil and would be a better choice for cooking, whereas Olive oil might be better as a dressing or a dip.
Next, I tried to answer the question “How to increase fiber intake ?” or “What contains fiber?”, and this is what I got from the system –
From the Bewgle analysis, I could get a snapshot of what to do and not to do on the path of lowering cholesterol naturally. I could also get answers to open-ended questions like “What contains fiber?”. I was pleasantly surprised about how the sentiments directly correlated to what’s good or bad for cholesterol.
Keeping aside the tech, on a personal note, for people who are interested, these are the changes I incorporated, which gave me significant results:
- Added at least 30 min exercise / walking / physical activity routine.
- Started having garlic and lemon with lukewarm water in the morning on an empty stomach. The blogs that I added for the analysis here do not cover this, but this combination helps in detoxification of the liver.
- Switched from toned milk to skim milk. (Pay special attention to dairy intake. Read the labels carefully. This can be a significant portion of the saturated fats in your diet)
- Added fiber to my diet in terms of oatmeal, fruits and beans.
- Reduced sugar and butter intake.
- Removed white rice / flour intake. Instead added red / brown rice or millet based flour.
- Added nuts and seeds to my diet. Now, my evening snacks mostly consist of these.
- Reduced the simple carbs intake at night, instead added salads or millet based menu.
I don’t think there’s a single mantra for everyone, but in consultation with your doctor, it’s possible to incorporate natural changes to see if it makes a difference. As correctly detected by Bewgle, ‘Lifestyle Changes’ is one of the most important factors in this journey. While you may not see changes in a few days or weeks, with a healthier lifestyle, hopefully you will see changes in about 6 months time as I did.
Though I limited this analysis to blogs, we can easily add Youtube content to this analysis as well.
If you have a use case or would like to try Bewgle on your unstructured data, please reach out to us at firstname.lastname@example.org