Identity Verified Thinker in Business / Industries / Pharmaceuticals
Dennis Lendrem
Dennis Lendrem
Making Data Make Sense
Aug. 23, 2017 6:30 am
Automation bias is the tendency to favour suggestions from automated decision-making systems and to ignore contradictory information made without automation, even if it is correct. And with the growing dependence upon automated methods, it's a problem in laboratories worldwide. To err is human. But to really foul things up, you need a computer. Paul Ehrlich You may...  Read More
May 7, 2017 4:54 am
In science, size matters. As a rule-of-thumb, the power to detect true differences - in experimental treatments, between disease groups, in the control and drug arms of a clinical trial - is greater in larger studies. The larger the study the more likely we are to pick up differences larger than can be simply attributable to chance. This is more or less true. Of course, it isn't totally true. ...  Read More
Apr. 23, 2017 4:09 am
Yesterday, April 22, 2017 scientists marched for science. At rallies in capital cities across the world they took to the streets to support evidence-based policy and reject policy-based evidence. And it was a remarkable event. Because this kind of thing does not happen in science. From time to time, individual scientists - Linus Pauling, Albert Einstein, Carl Sagan - have...  Read More
Apr. 1, 2017 2:59 am
You can lead a man to the light, but you can't make him think. Norman Einstein, CEO Scientific Radicals Multiple regression analysis is a really useful technique. It allows us to model the simultaneous impact of several variables on an outcome of interest. But in the 1980s, multiple regression analysis developed something of a bad name. The advent of modern computing meant that...  Read More
Mar. 26, 2017 5:54 am
Heads up everyone. The following 'dodge' is wrong. When analyzing data, a standard 'dodge' is to turn a continuous measurement into a binary, all-or-nothing variable. For example, we can define a cut-off point for cholesterol levels where values above the cut-off are defined as 'high' and values below may be defined as 'normal'. At first glance, it appears sensible. We have clearly defined...  Read More
Mar. 18, 2017 3:36 am
I was at a one-day meeting on "The Development Pathway for New Medical Technologies" hosted by the NIHR Diagnostic Evaluation Co-operative at Newcastle University . One of the exercises was to map "Real-World" experiences to the development pathway. And, once again, the group traced the root cause of failures in the development life cycle for new medical devices to misconceptions at the...  Read More
Feb. 19, 2017 4:10 am
I was wandering about the laboratory on Friday. I stopped, looked around the lab and thought 'You know what? This is a pretty international crew!" At the benches about me were a Japanese scientist, two Chinese, two Spanish, two Italians, two Greeks, a Pakistani, one Irish, an American, and three Brits. The banter between us was in English. And it was funny. And, in truth, its always been like...  Read More
Jan. 3, 2017 7:46 am
One of the challenges when modelling biological and other complex systems is that the resulting models often work fine until they are tested in a dataset other than that used for the modelling. This is the Catwalk Effect. While the models often fit really well and look quite stunning, they do not translate beyond the catwalk of the scientific journals. They may be sexy – they look great and...  Read More
Dec. 5, 2016 5:25 am
So, here’s the problem. I want to know if my patients are responding to treatment. Those that are? These are my responders. All well and good. Those that aren’t? These are my non-responders. Maybe I ought to change their treatment? I'm interested in what separates the responders and the non-responders. I divide my patients into responders and non-responders. I then carry out a ton of...  Read More
Nov. 17, 2016 6:50 am
It’s funny coming back to BestThinking. I’ve been away for a while but I had to share something with you. You know how, sometimes, the stuff around us becomes so familiar we don’t see it any more? This week my eyes were opened by Prof Sarah Teichmann. She showed us something that I’ve looked at a thousand times before but never seen. I've reproduced the slide below. The image we're interested...  Read More
Oct. 12, 2016 9:11 am
Skelton's Law states: "If you're going to get it wrong, you might as well go about it the right way" It first arose in the context of the internal validation of prognostic and diagnostic models. Internal validation checks - sampling, bootstrapping, cross-validation - give little or no assurance that a model will work in real life. The Counsel of Despair runs "It's never going to work so...  Read More
Jul. 29, 2016 4:13 pm
I abandoned filing my email about twenty years ago. I learned that filing email made finding email more, not less, difficult. And, with the advent of modern search tools, I didn't need to file emails in order to retrieve them. Creating folders made retrieving email more difficult. Especially if I dragged and dropped to the wrong folder! I've been called upon to defend my strategy on...  Read More
Apr. 9, 2016 9:02 am
So, here’s the thing. The scientific literature is awash with studies identifying biological differences as potential biomarkers. The problem is, that you can have a very profound biological difference between two groups of patients, but that difference is profoundly useless as a biomarker. Say what? To illustrate how this paradox arises, think of a well-established biological difference –...  Read More
Feb. 18, 2016 6:13 am
I have had my results for a long time: but I do not yet know how I am to arrive at them. Carl Gauss (1777-1855) Sometimes, when working on a problem, I may intuitively know the answer. But it can take many months before I work out the solution and can show my working. I'd like to think that this is because I'm a genius. But I'm not. The reality is that I often think in...  Read More
Feb. 7, 2016 4:50 am
The other day, I was faced with a complex dataset for analysis. A simplistic analysis presented massive multiplicity problems: we had multiple cytokines, at multiple time points, at multiple doses and multiple treatments. Initial plots of the data suggested there was not much going on. However, a multivariate analysis of variance flagged some real differences. Interpreting those differences...  Read More
Jan. 31, 2016 4:44 am
Ignore outliers at your peril: Outliers are often the most valuable data you have. Outliers often fly in the face of the scientific narrative. While, other studies have reported that identifying data that are inconsistent with current thinking is one of the critical success factors leading to scientific breakthroughs 7 , we observe that many jobbing scientists seek to...  Read More
Jan. 24, 2016 4:09 am
Stories help people make sense of the world. But when someone begins to believe those stories, and accommodate facts into those stories, they may become guilty of narrative bias. Theories help scientists make sense of the world. But when scientists begin to believe those theories, and accommodate facts into those theories, they may become guilty of narrative bias. Fortunately, scientists...  Read More
Jan. 6, 2016 12:31 pm
The Herschel Building in Newcastle University is home to a cluster of exciting, new,innovative spin-out companies at the leading edge of biomedical technology. I popped in there the other day to meet a client. As expected I learned that the building is named after the famous astronomer and composer William Herschel (1738-1822). Herschel is most famous for his discovery of Uranus in 1781. He...  Read More
Dec. 25, 2015 9:51 am
England is mired in history and tradition - "ritual's footsteps ankle deep in stone." One of the upsides of this is that with a little research, a trip to even the smallest and most obscure of villages or towns often yields some interesting historical surprises. Last weekend I found myself visiting the town of Burnopfield. If you've heard of Burnopfield I'd be rather surprised. It is a tiny...  Read More
Dec. 20, 2015 3:39 am
Every scientist knows that small, underpowered, studies are less likely to lead to statistically significant results. The Winner's Curse though is less well known. This refers to the observation that studies demonstrating an effect often provide inflated estimates of that effect. Such inflation is expected when an effect has to pass a certain threshold — such as reaching statistical...  Read More
Dec. 12, 2015 4:51 am
The website 'Shooting Tracker' collates data on mass shootings in the US. Mass shootings are defined as those with four or more victims. The US averages around one such mass shooting a day. I stumbled upon this analysis of mass shootings for the period 2013-2015: Honestly? I can't tell you how bizarre it is for people living in Europe...  Read More
Dec. 2, 2015 6:21 am
I've been looking at outliers and was reminded of Maier's Law. If the facts do not conform to the theory, they must be disposed of. N.R. Maier, "American Psychologist", March 1960 Thanks to Tim Everitt for pointing me in the direction of this paper.  Read More
Nov. 26, 2015 6:28 am
I met with those clever people at Lazy Grace last week. They reminded me again of the story of Abraham Wald . Abraham Wald was a brilliant statistician forced to flee Nazi persecution of the Jews in Hungary. Wald observed that the data we don't have are often the most revealing. Looking to add armour plates to protect returning bombers during the Second World War, Wald advised that the plates...  Read More
Nov. 24, 2015 5:13 am
Many scientists believe that small experiments, guided by scientific intuition, are simpler and more efficient than design of experiments. This belief is strong and persists even in the face of data demonstrating that it is clearly wrong. The Design of Experiments (DOE) allows scientists to identify those experiments maximizing the amount of information about what is happening in the often...  Read More
Nov. 22, 2015 4:21 am
This week's research meeting was really interesting. One of the problems with Big Data is there is tons of it, most of it is garbage, and usually you don't have the one bit you really need. We were running through the key results from the analysis of what is one of the largest data sets ever collected from patients with this particular disease. The interpretation raged back and forth. After ten...  Read More
Nov. 7, 2015 6:04 am
Some time ago I shared three blog items on "Why Do Scientists Usually Get The Answers They Expect?" Many scientists believe that small experiments, guided by scientific intuition, are simpler and more efficient than Design of Experiments. This belief is strong and persists even in the face of data demonstrating that it is clearly wrong. These have now been written up in the paper ...  Read More
Oct. 31, 2015 6:14 am
Any statement that equates two errors is wronger than wrong when one of the errors is clearly more wrong than the other. Asimov's Axiom As Asimov put it: "When people thought the earth was flat, they were wrong. When people thought the earth was spherical, they were wrong. But if you think that thinking the earth is spherical is just as wrong as thinking the earth is flat,...  Read More
Oct. 27, 2015 9:07 am
I've been reading a great little book by Siddhartha Mukherjee on The Laws of Medicine: Field Notes from an Uncertain Science [1]. There are just Three Laws. We covered the First Law in The Laws of Medicine: The First Law The Second Law states: 'Normals' teach us rules; 'outliers' teach us laws. The argument is that most data, most patients, most experience provide rules of...  Read More
Oct. 25, 2015 5:14 am
I’ve been reading a great little book by Siddhartha Mukherjee on The Laws of Medicine: Field Notes from an Uncertain Science [1]. There are just Three Laws. The first law may be poorly formulated, but states: A strong intuition is much more powerful than a weak test. While intuition is attractive, beguiling, and often wholly or completely wrong [2] – the thrust of the argument is...  Read More
Oct. 21, 2015 12:11 pm
Six months ago I was doing some calculations and came up with a crazy result. I thought, “That can’t be right.” I let it slide. I left it alone for a few weeks. But that crazy result bugged me. I went back to it. And played with it some more. No matter what I did, I couldn’t make it go away. I began to entertain a radical idea “Maybe it was right?” That meant a problem thought to be...  Read More
Return to Main Blog Page
Latest Ebooks