Sunday, January 19, 2014

This 15-Year Old Did Not Transform Medicine

I recently came across this news article about Jack Andraka, a 15-year-old boy from Maryland who invented a test to detect pancreatic cancer in its early stages. Naturally this is HUGE. For those who are not familiar, pancreatic cancer is one of the worst cancers one could get. Due to its lack of symptoms (as it's seated deep inside the abdomen), it's usually diagnosed at a very late stage, and the majority of people do not survive for longer than one year.

This simple, fast and cheap blood test Jack Andraka invented promises to change all that altogether, creating an unprecedented revolution in medicine by causing the greatest improvement in cancer medicine we have ever seen. This test costs just 3 cents, nearly 100% accurate, and won him the grand prize in the prestigious Intel International Science and Engineering Fair. The test uses nanotechnology to detect mesothelin, a type of protein which is found in the blood when one has pancreatic cancer. His wish is that this test will become widely available on the shelves of the supermarket, and everyone could just pick it up and do this test during their free time, and no one will die from late stage pancreatic cancer any more.

It's very nice, except that it does not work.

First of all, I would like to congratulate this bright young man for achieving so much at such a young age, and has dipped his feet into the world of scientific research and made a name for himself. To have your name on a "cancer sensor inventor" as a 15-year-old boy is simply amazing.

However, unfortunately that's where the achievement ends.

As a medical doctor I feel compelled to debunk the hype: This invention will unfortunately NOT save lives, and in fact I suspect if it were to be introduced as a 5-cent dipstick available in your local supermarket (which WILL NOT happen as you will see below), it may actually end up doing more harm than good to people's health.

It may be a difficult concept to explain but I hope you bear with me as I go through the reasoning.

I would begin by how making diagnosis works. It is often mistakenly thought that diagnosing a disease in the modern era is as easy as finding the correct protein in the blood, and BAM you have this disease. It's almost like if you find a fingerprint then BAM there has to be matching, unique person behind that fingerprint. However, the majority of medical diagnoses are simply not made this way.

I would use the pregnancy test as an example. We all know that urine or blood pregnancy tests are pretty accurate these days - it detects a hormone called βHCG which is secreted during pregnancy. So, if you find βHCG in urine or blood, then you are pregnant, right? WRONG. While the vast majority of high βHCG is due to pregnancy, sometimes it could also be due to sinister causes called gestational trophoblastic diseases which are a type of tumour in the genital organs. But in practice, if you missed your period and you are tested positive, then you would be told "you are pregnant" unless the doctor has deep suspicion that something amiss is going on.

This is because
  1. There are FAR MORE pregnant people than people with this tumour 
  2. The fact that you missed your period makes pregnancy even more likely.
βHCG is useful because:
  1. When it's level is very close to zero, then you can't be pregnant (It has good negative predictive value
  2. In pregnant people the level is ALWAYS elevated. (It is sensitive)
  3. When it's elevated, 99% of the time it's gonna be due to pregnancy (the other 1% being the gestational trophoblastic disease) (It is highly specific)
  4. When it's used, it enables good outcome (you know you are pregnant hence you commence antenatal care etc)
While these 4 conditions, especially the last, may seem trivial, they are THE criteria that any diagnostic test have to meet prior to being practical. If someone comes along and develop a 5-cent new pregnancy test, they will either have to meet these criteria, or being dumped despite being only 5 cents.

That's for diagnostic test. Moving on to screening test. Wouldn't it be nice if we find a test for early stages of various cancers, so that all we need to do is to wake up everyday and dip a few drops of blood, and we would know that we have (or not have) cancer? Yes it would be nice, but unfortunately medicine is hard and nothing like this exists, and no, Jack Andraka's dipstick is not the elusive magic test.

I would use PSA as an example. PSA (Prostate-Specific Antigen), as the name suggests, is a protein quite specific to the prostate, and is elevated in prostatic cancer. We used to do PSA screening quite commonly to detect early prostate cancer (but it's no longer recommended but that's a long story on its own). The problem with PSA, as with many other types of cancer blood tests, are that they are not specific and often not sensitive enough. In PSA's case, there are many other conditions which also increase its level (namely large prostate, severe infection etc). And last but not least, because prostate cancer is such a slow growing tumour, it's been found that even after using PSA and detecting some earlier cases, the mortality rate (chance of dying) is THE SAME whether or not you test everyone for  PSA. Hence population-wide prostate cancer screening is no longer recommended.

Moving on to mesothelin and pancreatic cancer. 

For the scholarly minded, this is THE article that shows why mesothelin is useless as a pancreatic cancer screening marker: 

Jack Andraka is right in pointing out that mesothelin is almost always present in patients with pancreatic cancer. However, mesothelin is ALSO present in ovarian and pleural cancer, AND in normal healthy people. The range of mesothelin level amongst pancreatic cancer sufferers overlaps greatly with the level amongst normal population. Even though Jack claims this to be 100% sensitive, it only means that it will detect a particular level of mesothelin 100% of the time. It still does not meet these criteria:

1. Does mesothelin differentiate between different cancers? No as it's also present in ovarian and pleural cancer. 

2. Does mesothelin differentiate between disease and health? No, when you are "positive" for mesothelin you may very well be healthy. 

In other words, if you bought this test and is tested positive, you could either have pancreatic cancer, other pancreatic conditions, ovarian cancer, pleural cancer, or have nothing at all. Not that useful isn't it? 

At this juncture, some people might claim, even if some healthy people mistakenly test positive in this test, they could always just do more tests and find out that they don't have disease - isn't that better than the alternative, having pancreatic cancer and not knowing it? The answer is NO. As pancreatic cancer is such a rare disease, you will have far less disease detection rate (true positive) than false positives. The thousands and thousands of people who had false positive results will now have to go through more tests (CT scans, biopsies etc), and all these tests actually do harm if you are healthy (CT increases your risk of cancer, biopsies are invasive procedures and put you at risk of infection and bleeding). So in the end, having such a test, despite costing only 3 cents, will end up putting a lot more healthy people at risk of complications of over-investigation than saving a few lives from its actual detections. 

To sum it up: Yes this man has a bright future, but he's not a cancer saviour, and we still have a long road ahead in our battle with cancer. He is not the genius kid who managed to discover something that millions of scientists in thousands of universities have overlooked in decades of cancer research. Unfortunately there has been a huge media circus surrounding his invention, most of which were more focused on perpetuating the "prodigy cancer saviour" feel-good story without getting an established scientist to put things into context. The whole media circus has planted a distorted perception on cancer research, and could end up instilling distrust amongst public in proper scientists and researchers. In the comment section of the aforementioned news article, the top comment is about how such an invention (like the many dozens of "cancer cures" invented each month) will never see the light of the day because pharmaceutical companies need to keep making money from cancer treatment drugs rather than saving people's life with cheap, easy and effective inventions like this. It insults the efforts of millions of scientists in labs everywhere, who toil away in their often frustrating and mundane efforts day in day out, without the benefit of being glorified in the media as a cancer saviour.

When something sounds too good to be true, often it's because it is too good to be true.

Footnote: This article by Forbes Science is one of the rare media articles which summarised the hype surrounding Jack Andraka instead of joining the media circus of how we have found the young saviour which will save millions of lives, before he even published a single journal article on the invention.

Footnote 2: Another article with a LOT more details about doubts on Jack Andraka's invention and personal motives, though it is a lot more sensational and may sound more personal. 


Monday, April 30, 2012

6 ÷ 2 (1+2) = ?

Personal Note: It's been more than a year since I last wrote on this blog, and let's just begin by stating that, YES I AM ALIVE. It's been an interesting fifteen months, where I transformed from a sheepish new intern with a quavering voice while saying "Hi it's Yang one of the... doctors" to a more confident doctor who is still humbled everyday by the nature of the amazing job. Oh and I also got engaged in the process. :)

You have seen it before. Yes, this is a reincarnation of the infamous 48/2(9+3). 

If you have not seen it, take a few seconds to work out the answer. In fact, even if you have seen it, try to solve the equation again in your head. 

Now that you have got it, let's check the answer. The "correct" answers are 288 and 9 respectively. Now, I will explain the quotation marks in a second.

When the original 48/2(9+3) question was released into the World Wide Web, it cooked up a storm as people hotly debated whether the answer should be 288 or 2. The "correct" answer is derived based on the strict interpretation of the "BODMAS" rule, which stands for Brackets, Order (Exponent), Division and Multiplication, Addition and Subtraction. Based on this rule, the operation should be
48 / 2 (9+3)
= 48 / 2 (12)
= 24 (12)
= 288
On the other hand, the proponent of the answer of 2 works it out this way:
48 / 2 (9+3)
= 48 / 2 (12)
= 48 / 24
= 2
Evidences have been thrown about in support of each argument. WolframAlpha and Google's default calculators both give the "correct" answer of 288. [1][2] However, different scientific calculators give either versions of the answer depending on brands and models. 

So which answer do I think is correct? I think the first one is technically correct, but the second one is not wrong either. The biggest mistake is in fact the person who wrote such an ambiguous expression in the first place. 

Before we go any further, let me introduce you to this video by vihart (starting from 2:32)

"I would like some juice or water with ice - do you mean you want either juice with no ice or water with ice, or do you mean you want either juice with ice, or water with ice?"

Essentially this argument is pointless. It detracts from the true spirit of mathematics which is to derive and discern fascinating pattern and relationship in nature based on a set of axioms. All this argument does is to delve into syntax which evolved arbitrarily in the evolution of mathematical notation  - it has NOTHING to do with whether the maths is right or wrong.

The conflict comes from the fact that "BODMAS" is taught in primary schools when we still use the sign "X" to mean multiplication, and the sign ÷ to mean division, and if you wrote out this equation 6 ÷ 2 X (1 + 2), then no one would have gotten it "wrong" based on the simple BODMAS rule. 

However, as we progress in our mathematical education, X is replaced by simply having two entities written next to each other, and ÷ is replaced by writing out the expression as a fraction. Because of this, when first presented with this writing of 6 ÷ 2(1+2) [also note the very intentionally misleading spacing in the original photo], the intuition in anyone who have become familiar with the advanced mathematics notation would automatically translate this into:
especially due to our natural instinct of grouping the multiplication together when a bracket is involved. So this is where the mistake came from. 

Just throwing a last question here to illustrate my point: what is 1/2x? Is it half of x, or is it the inverse of 2x? The answer is, it is neither, it's just a poorly constructed mathematical expression, and the debate on semantics is just a waste of time.


Friday, May 06, 2011

On The Celebration of Death of Osama bin Laden

Osama bin Laden is dead.

I have been reflecting about the crowd's jubilation and celebration over his death, and how ethicists are at a dilemma of whether or how to justify our reactions. I think the explanation comes down to the very core of human nature - we are still animals regardless of how we would like to think of our established civility and morality.

The reason war, schadenfreude, and revenge are so much easier to do, and I dare say prevalent than love, forgiveness and peace-keeping, is that we are hard-wired to behave this way. Even though we were taught to believe in "人之初性本善" (the human nature starts out with kindness), I don't think we need to teach a kid how to revenge a bully, whereas it takes a lot more effort to inculcate the value of forgiveness. In neuroscience we have the fight-or-flight circuitry built in as part of the basic neuro-circuitry, but there's no such anatomical equivalent for love. I am not a neuroscience expert, but from what I could deduce, morality, love and forgiveness remain a higher-level, cerebral level of mental function, whereas hedonism, aggression, animosity towards enemy, herd mentality etc remain the domain of more primitive human behaviourism.

In the case of OBL, I think a lot of people hold such profound hatred towards him that the primitive reaction has overwhelmed the moral code of not celebrating a person's death. I don't really think too negatively of the crowd reaction, I think it reflects on just what we really are. Ethicists are fumbling over things that are really outside their domain. If someone revealed a national security secret after waterboard interrogation by his enemy, would you blame him for being unpatriotic?

Image credit: FBI - Ten Most Wanted


Saturday, January 01, 2011

[Photo] A New Year

A Moon Gate

A new year, just like a moon gate, is a separator we create to demarcate the future from the past. By decorating it we tempt ourselves with a promise that what's ahead is better than what's behind.

Happy 2011, may you have a fruitful year with many good returns!


Friday, December 24, 2010

Merry Christmas!

1/20s . f/4.0 . ISO 800 . 17 mm
The Boulevard, Ivanhoe-ho-ho

Kids watch a decorated lawn during a Christmas Light event in Ivanhoe, just outside Melbourne. For 60 years, the residents of Ivanhoe's Boulevard and surrounding streets have been decorating their houses and gardens with vibrant Christmas lights. Commonly recognised as one of the most impressive Christmas displays throughout Melbourne, many residents make an annual visit to the area during the festival.


Sunday, October 10, 2010

Painting with Light (2)

Continued from here.


In 2006, I joined Fotoholics which is a photography club in Melbourne University. Over time my knowledge and experience grew, and I began to appreciate that the "photo" in photography doesn't just begin and end with a photo's brightness. The quantity of light obviously matters in a picture, otherwise you would be looking at a black canvas. However, at the end of the day, it is the quality of the light that delivers a far greater impact on the audience.


Through activities like workshops and outings, I learned a lot about the quality of light. Here we have the Fotoholics adviser Shayen demonstrating lighting in a studio workshop. Have you ever wondered why studios use so many huge light boxes for their models? No, it's not because they need lots of light, though these equipments could indeed be very bright. The answer simply lies within the quality of the light we could achieve with them. A big light source like the one we see here is used to achieve a soft, glowing quality as it avoids the casting of harsh shadow.


It's a difference between this...

Profile Picture (by changyang1230)

and this.


We don't really have to splash thousands of dollars to get studio equipments to get nice looking lighting. For example, this shot was captured in a safari bus without any special equipment. The safari bus was trekking through a rather boring part of the safari so I didn't pay much attention outside. I happened to notice that the light from outside the window fell on this girl evenly, so I turned my camera towards her and snapped while she was looking sideway. By using window light alone I got myself studio-worthy lighting.


The same principle applies when shooting things like flower. I used to shoot things under direct sunlight thinking that it would be "bright" enough, but the pictures shot this way all turned out to be rather uninspiring as the harsh direct light produces a lot of distracting shadows. So I learned to shoot in the shade - yes it might be darker and you might have to use higher sensitivity setting (ISO) or use a tripod, but the result would also be much nicer. The reason is that in the shade, all the lights are coming from all directions instead of just one spot in the sky, so the light would have a soft quality to it.


Using a camera on its automatic setting is fine in many cases, but in some cases the camera would get it wrong. Here Grace is sitting in a hall with lots of light filtering through the window panes behind her. The camera incorrectly interprets the scene as being too bright, so it tries to darken the picture. However, by doing so Grace's face now turned out quite dark as it is not as bright as the windows to begin with. It's Grace's face I am interested in, not the window panes, so the auto mode gave me a crappy picture.

Grace Looking Sophisticated

To fix the problem, I added some flash to the picture so that Grace's face lights up a bit. This, together with the coffee, makes her happy and shiny. Note that in this picture, Grace's face also has a soft quality to it, that is because instead of pointing my flash directly at her, I pointed the flash at the ceiling and let it reflect back towards her. This technique, called "bounce flash" in photography jargon, makes flash light softer and is one of the most commonly used techniques in flash photography.


Nature is the best painter, but it is quite an unpredictable one at that. I was strolling along the Circular Quay, breathing in the fresh air of the late afternoon when I saw the Sydney Harbour Bridge glow under the golden sunlight. This picture was straight out of the camera and it had a fantastic colour to it. However, I was a bit unhappy with the little pole smack in the middle of the picture, so I walked around it...


... and took this shot merely 3 minutes later. The sky was still a very pretty violet, but the golden sunlight was gone. You can see here just how important it is to be at the right place at the right time, and more importantly, to be able to recognise wonderful light which is all too often very ephemeral.


Photographers often talk about the golden hours of the day, i.e. the few hours around sunrise and sunset. In these hours not only nature showers us with a whole gamut of glorious colours, it also delivers them in a soft, diffused and glowing way. I once read somewhere that an editor in a renowned landscape photography magazine has a policy of only accepting pictures taken during golden hours.


The landscape exudes an incredible aura during the golden hours. The next time you come across a wonderful landscape photograph, pay some attention to the time it is captured. More often than not they would be shot during sunrise or sunset, or even during dawn or dusk. That editor may have a point after all.

Sunrise at Grampians (Uncropped)

This picture was taken in Grampians National Park which is about 4 hours away from Melbourne. The difficult thing about chasing the golden hour is that you either have to wake up at ungodly hours, or continue shooting late in the day when everyone else would have had enough for the day and had started thinking about dinner. For this picture, I woke up at 5 so that I could be there just when the sun was rising. I got lucky, the sun shone through a gap between the clouds and gave me this surreal scene.


Of course the golden hour is not limited to landscape alone. It worked equally well in this cityscape of Shanghai.


You may not know that our eyes are actually much better at perceiving light in most regards compared to even the best camera in the world. This is your standard postcard picture of the two most iconic structures of Sydney, taken from the best vantage point during the best time of the day. But I cheated here. This is actually not just one picture. What you are seeing is actually a combination of three pictures.


1. This is what you would get if you just shoot it without changing any setting on your camera. Note that the Opera House is a bit dim here.


2. This is what you get when you take the picture of the same scene, but exposing for longer to get a brighter version. Notice how the Opera House is now bright, but the sky is so bright that you can't even see the details of the clouds anymore.


3. Lastly this is the dark version of the same picture, and it allows us to see the detail of the sky.

The three pictures of the same scene with different brightness are then combined to give us the final picture. The final picture is the most similar to what you would see with naked eyes, however due to the camera's limitation of the range of brightness it could capture, if you only took one shot you would not be able to get the same result. The technique of combining pictures of different brightness is called "High Dynamic Resolution" (HDR). The Grampians shot above was done with HDR too.


Another HDR shot: Sydney skyline.


I am still in the process of appreciating the quality of light. Be it contrasty light,


straight-from-above light,


selective illumination


or bright background light ("high key"), it's the way we paint the picture with light that counts the most in the final outcome.

And I will keep on enjoying light-painting for years to come.


Wednesday, September 22, 2010



该找些事来做啦。 如果以后还是在异乡过中秋一定要提一提灯笼,吃一吃月饼。虽说我不是最传统的人,但在这样的日子里独处的时候,我还是不禁感受到那团圆、团聚的意义,还有明月寄相思的韵味。