Machine learning is a process where computers are used to analyse past data in the hope of predicting outcomes from future data. These outcomes can be anything in the modern world of Machine Learning, ranging from what type of book or CD you are likely to buy to predicting the outcomes of sports events such as horse racing. Various different Machine Learning methods have been devised each with different strengths that can make them more applicable to different types of problems. So a Neural Network may perform better on one kind of problem than say a Random Forest algorithm, but less efficiently on another kind of problem. Risk factor calculators that you plug your data into on the web are based on simpler models which assume a linear relationship between the factors eg LDL, blood pressure etc. Machine learning algorithms can dig deeper so to speak and amongst other things, uncover weightings to various factors. Figure out which are most important and weight them accordingly.
Such an approach was taken with a 10 year project tracking people with 48 factors. Four different algorithms were employed on the data. The data was split into 75% to find out what the relationships were within the data, usually called ‘training’ the model in machine learning parlance. The remaining 25% was used to test how well the model could predict cardio events such as heart attacks. The results were pretty good, out performing conventional risk assessors. Taking an average of how the four methods ranked the different factors we can see that LDL is well behind HDL, Trig’s and HbA1c as a risk factor. Here is an ordered table of those averages showing that age (not surprisingly) was the most impactful feature. Note that some are negatively significant eg Women are significantly less likely to have an event). ‘Missing’ means within the data a patient had this data missing. here is a link to the report
Ethnicitya: South Asian
SESb: 2nd Townsend quintile
SESb: 3rd Townsend quintile
SESb: 4th Townsend quintile
Oral corticosteroid prescribed
Systolic blood pressure*
SESb: 5th Townsend quintile (most deprived)
Family history of CHD < 60 years
AST/ALT ratio missing
Ethnicitya: Chinese/East Asian
Chronic kidney disease
Anti-psychotic drug prescribed
Severe mental illness
Blood pressure treatment*
gamma GT missing
Serum creatinine missing
Serum fibrinogen missing
LDL cholesterol missing
During February I had a reminder of the dangers of over consumption of simple Carb’s. I spent a week in Rome and if you have been there you will know its wall to wall Pasta and Pizza. After a couple of days I just gave up and decided to eat the stuff for the remaining 5 days. I normally never touch Pizza and Pasta but made an exception in Rome. Now I normally spend a few months away in the winter and return to the UK exactly the same weight courtesy of the excellent dietary habits of the Portugese. A week in Rome however found me 7lbs overweight. This was a stark reminder of how damaging these products are.
Perhaps on a more interesting note I have just stumbled upon an inflamatory marker that has been shown to be a very good predictor of future or current heart disease. It is called the Neutrophil to Lymphocyte ratio and has been show to improve the Framingham risk score for any one who believes in the Framingham score (probably the people from Framinghan do). As usual this test can be obtained cheaply in Portugal for 6 euros. More details on the ratio can be obtained from the link below.
Just got my NLR results back. My Neurtrophil reading is 3.57 whilst my Lymphocyte reading is 2.08 giving a ratio of 3.57 / 2.08 = 1.71. The acceptable range is 0.78 to 3.53. Higher than 3.53 would be considered a problematic reading.
Higher HDL levels are generally associated with decreased risk for heart disease because of its role in transporting cholesterol away from macrophages within the artery wall back to the liver for recycling. However it seems that the amount of HDL is not the primary driver of improvement in this area. HDL Efflux appears to be the main driver and refers to how good HDL is at ‘sucking’ up cholesterol. You can have HDL C levels above 2.0 mmol but if your Efflux is low the hoover just ain’t picking up the bits. This may explain why very high levels of HDL dont seem to be better than just above average levels.
How can we improve HDL Efflux, well it seems the good old Med’ diet comes to fore once again along with a moderate (glass with a meal) amount of red wine. More controversially for me and other more expert in the field is the advocacy of extra virgin olive oil. Nothing seems to split the heart health community like this ingredient. Some research swear by it whilst others clearly show it creates endothelial disfunction. My feelings are that when confronted with split opinions like this I prefer to leave it out of my diet especially when its non essential. I will stick with a med’ based diet and a handful of nuts each day.
You may have come across the talks by Ivor Cummings on cholesterol. Ivor is a chemical engineer by trade and after receiving some personal alarming blood readings he decided to drill down into the meaning of what his readings actually meant. Applying his engineering rigour to the problem he quickly discovered that the emperor was wearing no clothes or the doctor in this case. I came across another engineer Dave Feldman, that has also gone into even more detail to extrapolate the real relationships between what we eat and how it effects cholesterol and associated lipid readings. It is a real beauty of a presentation which if you are new to the subject might take more than one listening. You can also easily replicate the experiment yourself, in fact he is appealing for people to try it and log their results on his blog.
The two things that have exploded in consumption along side the widespread development of heart disease are the consumption of these two food sources. We were all advised by the diet guideline experts in the 70’s to switch from saturated fat to vegetable oils and veg’ oil even today is common place in UK kitchens. Restaurants not only use it but to save money only change it when its oxidized to exhaustion. This study, until recently kept hidden, makes you wonder how we ever got on this path. It also suggests that what we eat is a bigger threat than cholesterol per se.
During the Vietnam war the US realized that they could create more damage to the opposition forces by inventing a bullet that would maim opposing soldiers rather than kill them. Why was this and what does it have to do with the staggering increase in Autism we have experience in just one generation. Mercola talks this interview up in the introduction but he is not hyping, its simply a must watch, pass it on to others
I recently had an exchange with Dr Mark Porter. I took exception to his recent article in the Times which more or less trashed the idea of people using Vitamin supp’s to combat disease. I challenged this firstly on the Vit C front (but also on others) and I am glad I did because you always learn something when you talk to people with opposing views. It gets you off your backside to investigate. Dr Porter kindly replied and sent me a link to a meta analysis that essentially showed that no Vitamin was beneficial in a statistically significant way in relation to heart disease amongst other diseases, when supplemented, although selenium came pretty close. A meta analysis by the way is when a researcher takes all the research papers done on a topic and then pools together all the results to form one overall result. It also enables them to filter out badly designed studies. I decided to dig a little deeper into the meta analysis and in particular the heart disease and Vit C category. I am convinced that Vit C can impact Lp(a) and hence CVD. When I looked closely at the 7 studies forming the meta analysis on Vit C I noticed that they had weighted them. Now I forget the exact method of weighting but it sure wasn’t based on dosage because one study was weighted with 78%!! of the overall value of the analysis and this study only used 250mg of Vit C per day on people with heart disease. This study came out as pretty much neutral on Vit C and given that the overall score OR was 0.99 for Vit C (the more below 1 the better) you can see that this one study had massive or should I say 78% effect on the overall outcome of the meta analysis. As I said to Dr Porter, Linus Pauling would have laughed at 250mg even for healthy subjects. Incidently getting back to Selenium there is an interesting Swedish study that showed in subjects over 60 that a combo of Selenium and Coq10 slashed risk by 50%.
Footnote – Just discovered that Dr Porter has recently adopted a similar attitude to Carb’s as myself
Over the last three decades total cholesterol has risen in japan but Heart disease has dropped. The linked report suggests that the reason for this may be their higher than average HDL and healthy Total to HDL ratios. The other interesting aspect of the report was the differences logged in Dietary intake between the Japanese, American and Australian subjects monitored. The main differences were
Fats significantly lower in Japanese
Fish and sea food much higher in Japanese
Eggs much higher in Japanese
Tofu and related products much higher in Japanese
Mushrooms much higher in Japanese
Fruits higher in Japanese
were pretty much the same across the three countries.
With soil being continually depleted of magnesium we are getting less and less in our food supply. The research list connecting Magnesium deficiency and heart disease is long. here are just a few
Lazy solution – Eat a handful of almonds and or pumpkin seeds with your breakfast
I have mentioned before that pretty much all my ratios and markers are good eg Total to HDL ratio, ApoB/Apo A ratio, Lp(a), Homocysteine. One marker that I have found on two occasions to be too high is Lp-PLA2. This is a marker that is believed to signify unstable plaque within the arteries. So far I have not managed to lower it. I have however just got the results back for another test that looks at oxidative stress and a predictor of unstable plaque, namely Myeloperoxidase. I was kind of expecting this one might come back on the high side given that it appears to be a close cousin of Lp-PLA2 or at least testing a similar condition. I was very happy therefore to find that my reading was super low. Greater than 5 IU/ml is considered positive but I came back at less than 0.01.
The test was done with Blue Horizon and cost around £95. I will retest Lp_PLA2 later in the year and if it has not diminished perhaps try a spell on aspirin and see if that puts a dent in it or perhaps take solace from this test, surely they can’t both be right.
Good explanation here of the difference between Myeloperoxidase and Lp-PLA2
It would appear that the obvious approach to lowering Lp-PLA2 would be Niacin (Vit B3)