Criminal Chemistry: What's Your Poison?

How do you teach a machine to listen, and why would you want to do so? Georgia Mills broke the ice, or in this case glass, with Chris Mitchell from the Cambridgeshire-based company Audio Analytic...

Chris - Audio Analytic does a completely new type of sound signal processing. We call it audio artificial intelligence and it looks at giving computers the ability to respond to sounds such a baby crying, or glass breaking as opposed to speech or music type signals. One of the shortest descriptions I’ve seen of what we do we’re like the Shazam for real world sounds which, I think, is a reasonably short way of describing what we do. It’s technically highly inaccurate but it’s very short.

Georgia - How are you teaching a machine to tell apart sounds?

Chris - First of all you need to get the sounds. One of the fundamental problems we have is what’s referred to as a zero data problem, which means there is no data to train a machine off. However clever AI is, it’s not clever at all if you don’t have data to train off in the first place so we had to go out and collect the data. That means literally smashing, crashing, and bashing, and beeping our way through all of the sounds that you can possibly imagine. Once we’ve collected all the data you’ve then got to go through and label it.

Then you move onto the more AI side of it, which is then trying to model these sounds and the combination and ordering in which they occur. Now that is very different in itself from speech in the fact that there’s no language associated with sounds, so you and I having this conversation is governed by a relatively formal communication structure we both understand. We know the rules of it, around it and that means that at a sound analysis level you can guess which sounds are going to come where based on those rules. My dog doesn’t really have a sense of those rules in any meaningful sense so that doesn’t apply. A lot of that language technology and other aspects that you would approach from the speech side just simply don’t translate and, hence, you have to build a whole new set of AI to be able to accurately detect and respond to those sounds as they occur.

Georgia - So the process, vastly simplified, involves breaking the sounds down into tiny components and getting artificial intelligence to learn which components come from which sounds, and in what context. But I was most interested in the side of data collection, especially the sounds of smashing windows. Now a lot of this happened in a soundproof portakabin in Cambridge where I was shown their glass breaking rig…

Moray - My names Moray Grieve. I’m the data and QA manager of Audio Analytic.

Georgia - So I can see a sledge hammer in the back there - it looks very exciting. So I’m going to smash this window?

Moray - You’re going to smash it - correct. What we’ll do is we’ll get you kitted up in safety equipment. You’ll be behind the glass break rig so you will be as if you were a burglar breaking into the room. It may take you several attempts. It’s actually a lot harder than a lot of  people think. A lot of people think you can just tap the sledge hammer to the glass and it will break but, in reality, you do have to give a reasonable amount of force in order to break it.

Georgia - I’m so excited. I cannot tell you how excited I am swing a sledge hammer into a window. This must be the best place to work ever!

Moray - It’s great fun - it is. It’s all been a bit of an initiation for everyone when they start.

Georgia - Right. So I’d better get kitted up then.

I was wearing about five layers of protective clothing to prevent glass from shredding me, and ear defenders as it was incredibly loud. So, as fun as this was, not something to try again at home. And, believe it or not, it took seven attempts with a sledge hammer to get through a pane of the famously breakable material… glass.
But eventually… glass breaking.

Georgia - What’s the point of this learning to hear for machines?

Chris - There’s security and safety applications to what we do. If we’re selling into cameras that people store in the homes, detecting whether the windows are smashed and then sending you a text message alert is, obviously, extremely valuable in terms of being able to do that. Through to modern smart speakers and understanding the sound environment in which you project sounds. If you can have an understanding of that -  so that if you know you’re frying bacon, that has an impact if you’re trying to play jazz -  so understanding that broader scene from a sound is very important as well.

Georgia - Oh, I see. I have this problem when I’m trying to my podcasts while making dinner. So how would understanding the sounds help me, I suppose, listening to my podcast?

Chris - Speakers are generally, at the moment, are pretty dumb. If it could understand that you’re making bacon - I sound like I’m obsessed with bacon but fair enough! If we go back to the bacon piece knowing the sorts of sounds involved in that and inflecting the way the music is produced so that it has the best chance of getting over the top of that sound and around it, means that you’ll be able to enjoy your podcast while still cooking the bacon in this case.

Georgia - How accurate is this? I guess compared to humans because quite often, we mentioned babies crying, I’ve been spooked by what I thought was a baby and it turned out to be a fox in the garden screaming. Are there any sounds it’s confused by and how accurate is it?

Chris - It’s highly accurate in the metrics we use. There are, inevitably, things that will sound very similar to it  and will confuse humans if you want to use that as a metric, which some people will do. For example, I have an engineer who’s extremely good at reproducing a baby crying - very, very eerie. It’s taken him many years to be able to reproduce that so I don't know how much of a practical problem that is.

We’ve got a large deployment in Europe and we saw that there was a specific type of bird that’s kept in houses, especially in the south part of France, which sounds exactly like a North American smoke alarm. They are literally identical and the guys had to spend a bunch of effort getting rid of that to make sure they could be discerned. So you do come across these instances, but generally it’s highly accurate.

"Get fitter"...."lose weight"...."learn another language"....Most people approach New Year's resolutions with the idea that it will kickstart a permanent change in our behaviour, going from a slog to an automatic habit - what’s known as reaching a habit peak. But why do we make New Year’s resolutions in the first place and do they really work?  Izzie Clarke spoke to psychologist Benjamin Gardner from King’s College London...

Benjamin - There’s a popular myth out there that it takes 21 days to form a new habit. And I’ve looked into the source of this myth and it comes from a book that was written by a plastic surgeon in America in the 1960s in which he found that, among his patients, it took them an average of three weeks to get used to their new appearance. Now, as far as I know, that’s the sole evidence base for this claim that it takes 21 days to form a habit.

Around 10 years ago now, a colleague of mine at University College London conducted a study in which to study the habit formation process in humans. What she found was that 66 days was the average amount of time it took for people to reach their habit peak. Some people reached their peak in 18 days, other people didn't reach their peak at all in the 12 weeks. But, it was forcase that if they kept going, after 254 days they would reach their habit peaks. So there’s great variation between people, between different behaviours in regards to the length of time for things to start to feel like they’re becoming an ingrained part of our everyday routine.

Izzie - I have to ask: is there anything special about the 1st January in terms of changing this behaviour? Why do we even have this New Year’s resolution?

Benjamin - I think it’s helpful in one aspect in that it gives us something specific. We know that we are going to start our behaviour change attempt on 1st January and it’s a general principle of behaviour change that the more specific your plans are with regards to now only what you’re going to do, but in what situation you’re going to do it. The more specific our plans are in that respect, the more likely we are to stick to it.

But I think, on the other hand, I don’t think there’s anything particularly special about the 1st January. It’s useful to have a concrete date mind with regards to when you’re going to start changing your behaviour, but it doesn’t need to be the 1st January. You could equally set your behaviour change start date as say the 1st February or the 1st of March, or any day in the year.

In fact, I think because there is so much pressure to try and think of something you want to change in terms of your behaviour, I think it can backfire. I think people can have a go at changing their behaviour and be unsuccessful because their New Year’s resolutions often were unrealistic or they weren’t properly thought through.

Izzie - How important is having a plan for all of this?

Benjamin - It’s very important to have a plan because it’s well established within psychology that people often intend to make changes to their behaviour but, for whatever reason, they don’t get round to making those changes. One of the best ways that we can put our intentions into action is by forming a concrete plan specifying exactly what we’re going to do in exactly which situation we’re going to do it.

For example, if you want to lose weight, then you shouldn’t just set yourself a goal of losing weight, you need to think about what specific behaviours are you going to try and change in order to lose weight. It may be that you want to, for example, eat fruit snacks instead of eating unhealthy snacks while you’re watching TV. But then you can specify further to say well, what I’m going to do is I’m going to identify particular events where I snack, for example when I’m watching TV, and that’s where I’m going to substitute in the fruit for the unhealthy food that I eat at present. So having a plan is very useful because it helps to specify the details of exactly what we’re going to do. Just having a goal of changing your behaviour, or even just achieving some kind of health outcome is not going to be as useful as having a plan that tells you exactly what you’re going to do and in exactly which situation you’re going to go about doing it.

Izzie - Oh, I see. Because my New Year’s resolution was to get fitter, so you're saying that actually I can’t just be as broad as that, I have to have a real plan of break it down into the different components

Benjamin - Yes. I think if you’re plan is to get fitter then you need to think about which physical activities in particular are you going to do. Then having decided on that, and I’d argue that you should think that through very carefully, because you want to make sure that try doing physical activities that are realistic for you. Then you move to the next stage of saying okay, in which specific situations am I going to do those particular behaviours? So you might say to yourself, for example, I’m going to go to the gym more and when I go to the gym I’m going to use the running machine. In which case the next step is to say when exactly are you going to the gym? Are you going to go on your lunch break? Are you going to go after work? You need to think about exactly when you’re going to put these things into action, and the more specific you can be in that respect the better.

Izzie - I have to ask, do you have a New Year’s resolution for 2018?

Benjamin - I’m afraid I don’t. Perhaps I’m becoming more skeptical over time. I don't have any New Year’s resolutions. I also have two young children and I’ve found that the best formed plans can fall by the wayside when you have young children. So no, this year, I am not resolving to change my behaviour in any way.

How have poisons been used throughout history? Katie Haylor caught up with chemist and science writer John Emsley, author of Molecules of Murder and More Molecules of Murder, to find out...

Katie - As the old saying goes - “an apple a day keeps the doctor away.” But, eat enough apples in their entirety and you could, in theory, risk cyanide poisoning from all those pips. But now we know the ingredients of a good poison: low fatal quantities, being hard to detect, do we have any good, or should I say bad candidates?

The history books may be a good place to start looking. Chemist and author John Emsley knows his primitive poisons. He explains how the Victorians poison of choice was one readily available in the pharmacy, which meant some rules had to be put in place…

John - To purchase poisons from a pharmacist, you had to sign the poisons register and the pharmacists had to recognise who you were. You couldn’t just walk into any shop and buy it. That was done because there were too many murders mainly with something like arsenic, as it’s oxide was known as white arsenic and this was widely used as vermin poison and things like this. Colouring agents could be made from it so it was available. Because it was available it was being used to dispose of unwanted relatives and things like that.

Katie - Okay. Are there any particularly famous cases of this?

John - There’s some very famous cases in history. But going back even earlier to the 17th century the Marquise de Brinvilliers in France murdered her relatives to secure the family income. She was only detected because the man who was supplying it, her lover, suddenly died and they discovered her letters. The thing about that was this was one of the first cases in which someone was asked to prove that the powder they detected was the deadly poison. That was in 1676, and a pharmacist in Paris devised a set of test to prove that what they’d found was actually arsenic and it worked, and she was found guilty and executed.

Katie - What is it fundamentally about arsenic that makes it such an effective poison?

John - The think is it’s in the same group of the periodic table as phosphorus. Phosphates, of course, are essential in lots of parts of the human body from DNA upwards and, of course, if you put arsenic there the body can recognise arsenic and thinks it’s getting phosphates. It will introduce arsenic at all sorts of points and then, of course, these things won’t behave as they should behave and very quickly you’re made very ill. The body recognises arsenic as something it doesn’t want very quickly but not before it’s absorbed quite a lot of it. Then, of course, it screws up various parts of your body and, with a big enough dose, you die within a day or so.

It will attach itself to something called ATP (Adenosine Triphosphate) and, if does that, then it really interferes because ATP is needed in every cell of the body. Then things begin to break down because you need ATP to generate energy in your heart, for example. If you can’t generate enough energy then you’re going to be in real trouble.

Katie - The chemical element arsenic’s availability and toxicity made it the perfect poison of the day, but it could be detected. So could a poisoner try to mask their ill intentions by using a substance that is already found naturally in the body?

John - Adrenaline is one that has been used in murders. The chemical name for it is epinephrine, and this the body produces and it produces a lot of this if you’re under sudden stress.

Katie - This the ‘fight and flight’ response, isn’t it?

John - Yes. If you’re having a heart attack they may give you this to stimulate the heart and keep it going. It’s not something you would normally think of as being a poison but it has been used by nurses. Nurses, of course, have access to it and those are the ones who generally murder with adrenaline because they know if you give too much adrenaline it’s going to over-stimulate the heart and the heart will pack up.

The serial killer nurse called Kristen Gilbert probably killed about 50 of her patients with adrenaline. She worked at the Veterans Hospital in Northampton, Massachusetts and she did it for a variety of reasons. If the person began to annoy her, then she would give them an overdose. Often these people were very ill and so it was just put down to the person had died of natural causes. But, eventually, she got rather careless in that if it was someone in intensive care, and she wanted to go out with her boyfriend and the person in intensive care couldn’t be left, she would give them an overdose of adrenaline and pass them onto the next world. It’s not known exactly how many people she dealt with, but it was probably around 50 before she was finally caught.

The thing about that was it became so common this occurrence of people she was supposed to be looking after suddenly dying that they began to investigate. Other nurses said there’s something odd about this - I think they called her The Angel of Death, or something like that. They began then to analyse the body and, of course, were able to detect there was far more adrenaline in the body of the dead person than the body itself could actually have produced. They were able to go back and analyse samples from bodies that had been dead two or three years and there was still excess adrenaline.

Katie - Despite being a chemical that pre-exists in the body even adrenaline can’t stay hidden for long when given in lethal doses. So how about something so deadly, something which is needed in such a small quantity to cause harm, but it’s virtually undetectable?

John - The thing about polonium chloride is that it’s radioactive but it gives off only alpha particles. Alpha particles are relatively harmless; a sheet of paper will stop an alpha particle. It doesn’t give off gamma rays. Gamma rays are the things that you can use a geiger counter to detect. With polonium it was possible, the Russian Secret Police found, to pass through the parts of airports where you go through a thing and it will detect if there’s anything radioactive on you. They could pass through that so they were using this to dispose of people. The amount of polonium you need to kill someone is so tiny it’s difficult to see - it’s as small as a mote of dust. It is that deadly.

Katie - Polonium is radioactive. Once inside the body the alpha particles emitted by polonium can make their way into cells, damage DNA, and cause cell death.

John - We know our cells have got  a wonderful repair mechanism that’s being used all the time because cells are needing repair. Of course, it can be repaired but not when you’re giving off something like a 100 million of these particles a second, then there’s not way the body is going to be able to cope with that.

The most famous poisoning of this was someone called Alexander Litvinenko and he came to Britain. He was a KGB agent and after the end of the Soviet Union a lot of people from Russia could then come and live in Britain. He came to Britain and he was granted asylum because he said he was ex KGB and he was able to tell the British secret people a lot about it, so we took him in and he was given another name. He was called Edwin Redwalt Carter, and he was given a house with him and his family in Muswell Hill in London and he began to lead quite an anonymous life.
The trouble was, when he was visited by his wife’s mother, she went back to Russia and the police there knew who she was. They searched her luggage and they found out details of  who he was and where he lived, so they then sent another agent to London to dispose of him and that agent came with polonium chloride. He was an old “friend” of Litvinenko and he invited Litvinenko to the Millenium hotel where he was staying London, and there he had green tea with polonium chloride in it.

Of course, the following day he began to feel ill. The doctor diagnosed gastroenteritis, the usual reaction to something you’d eaten. Put him in hospital; he got no better; sent down to University College Hospital because he obviously was seriously ill. His hair began to fall out, which is sometime a sign of thallium poisoning, and thallium’s a deadly metal. But they detested for thallium and no it wasn’t thallium and so they knew his hair was falling out because of radiation damage. The chemists, the scientists, detected polonium and, suddenly, it was obvious what was killing Alexander Litvinenko, and he died after about three weeks in hospital.

Katie - Death by radiation. No taste, no smell, and given in miniscule quantities. It’s polonium’s volatility that led to its detection. Once identified, scientists were able to detect the traces it left behind on various surfaces, even including the seat in Arsenal Football Stadium where the killer had sat to watch a match.

John - We are only talking now of a few atoms and yet, there it is and there it can be detected. Even that supposedly undetectable poison that the Russians were using is now no longer in that category of undetectable.

Music: Daniel Birch & Ben Pegley

Poisons may be most associated with gruesome murders, but the vast majority of poison related deaths are accidental. According to the NHS website, most poisonings in the UK occur in the home. So are there compounds lurking our cupboards that we should be aware of? Katie Haylor spoke to Professor Robert Chilcott from the University of Hertfordshire...

Robert - If we stick to the consumer products, one of the more interesting ones in botulinum toxin. Sometimes, if food has not been processed properly, things which are stored and boiled in jars in tins can actually grow certain types of bacteria which produce botulinum toxin - botox.

Katie - That’s right, you heard correctly: botox is the same stuff that some people have injected into their faces to relax facial muscles in order to reduce the appearance of lines and wrinkles - it’s the same toxin.

Robert - It only counts for a very small fraction of percent of all reported food cases, but it’s really the one you don’t wish to have. It nearly always requires hospitalisation and about one in ten people who get botulinum toxin will actually die.

The medical form, of course, is very pure and it’s a specific form of botulinum toxin. In the wild there’s about seven types of botulinum toxin but they all have the desired effect; they’re incredibly toxic. I think botulinum toxin is the most toxic substance that we’re aware of.

Katie - So bacteria produce this toxin, and this bacteria could be around in the food that goes into the can, so what happens then? How, if you’re unlucky enough to eat this food, what happens to your body?

Robert - The bacteria that produce the toxin are pretty much ubiquitous. It grows nearly everywhere, so almost certainly, it will be in the food that you try and preserve. The problem comes when you don’t follow the right instructions. For example, if it’s not boiled properly or you keep it in a solution which is not sufficiently acidic, then you run the risk of the bacteria starting to produce the toxin, which is primarily under what’s called anaerobic conditions, so anything that doesn't’ have too much oxygen in will be a nice breeding ground for botulinum.

As I said, it’s an incredibly toxic material. The estimated lethal dose is about 70 nanograms. A nanogram is about a thousand millionth of a gram, so it’s very difficult to put that weight, or lack of weight, into perspective. But, as a rough guide, if you think about a single hair, that would be about 9,000 times heavier than a lethal dose of botulinum toxin.

Katie - Is this something that we should worry about as consumers of tinned foods, or does the problem come when you’re making your own tinned food?

Robert - I think the mass market, the processing now is to a point where we don’t need to worry about this at all. Of course, if we’d lived about 100 years ago when tinned food was just coming into fashion, that probably was more a game of Russian roulette. The most common forms of botulinum toxin these days is people who preserve the food at home: pickles and so forth and things which goes into jars. Fruit, in particular, because sometimes it lacks the acidity that you need to prevent the botulinum from growing. But, in general, it’s actually, as I’ve said, a very rare form of food poisoning.

Katie - Why is it so poisonous? What does this toxin do when it gets into the body?

Robert - To put it simply, the botulinum toxin works by turning off our nerves. As you probably know, nerves normally activate muscles using substances called neurotransmitters. These are chemicals that are released from the nerve endings and result in the coordinated movement of muscles, so when we speak or walk or breath, this is all down to the nervous control of our muscles.

Basically, the toxin acts by cutting off the ability of the nerve endings to produce the neurotransmitter. Essentially, that stops the muscle from being told to contract and, eventually, it just causes paralysis. What’s particularly nasty about botulinum toxin is that you get these paralysis effects and, eventually, you are unable to breath, but the whole time you're actually conscious and aware of what’s going on.

There is an antidote; it’s based on antibodies that actually bind to the different types of the toxins. There are seven different toxins in botulinum toxin. If you manage to get to a hospital and diagnosed then that antidote may be available. Otherwise, the only treatment really is supportive - you may need to have mechanical ventilation. But botulinum poisoning can last quite a long time so you may still suffer different degrees of paralysis for many months after you’ve initially been poisoned.

Katie - Wow. So if you’re thinking about making your own tinned foods you do need to be exceptionally careful. What about non-consumer products then; what about, for example, cleaning products can they be poisonous?

Robert - Oh, there’s a whole wealth of poisons in cleaning products. The obvious example may be chlorine. Most people have heard of chlorine, but it’s a bit of two-edged sword when it comes to our society. On the one hand you have your chlorine-containing consumer products that keep our world nice and clean and fresh and free from major diseases. But, of course, if you mix the wrong type of cleaning products together you can quickly produce lethal quantities of chlorine. It’s not uncommon: I think in the UK each year there’s about half a dozen major incidents involving public exposure to chlorine. There are probably a small number of incidents involving single people, people have mixed these chemicals in their toilet.

But chlorine is actually quite toxic. It’s a very highly reactive molecule and it likes to seek our moist areas, so when you breath it in the nice wet lining of your lung tends to absorb the chlorine. There’s no specific mechanism of action, it’s just nasty corrosive stuff. In a sufficient dose, you get the chlorine inside and it’ll basically just rot the lining of your lungs. That then causes a condition called pulmonary edema so, basically, all the fluid that’s inside our body starts to leak into the lungs and it stops us from breathing.

Katie - What kind of toxic combination are we talking about?

Robert - The rough rule of them is: don’t mix cleaning products. But the particular cocktail which has the worst effect is mixing a bleach solution with anything which is acidic or contains ammonia.

Katie - Household bleach contains sodium hypochlorite, a chlorine compound. It’s a common cleaning product often used as a disinfectant to clean toilets, swimming pools, and even to treat public water supplies. But, when mixed with other cleaning products, toxic gases can be released. For instance, chlorine gas in the case of mixing with acidic substances, or chloramine gas in the case of mixing with alkaline products like ammonia.

Robert - The initial signs of poisoning are relatively low concentrations, it’ll cause mild irritation of the eyes and the lungs. So, basically, it’ll be like going to a swimming pool for maybe an hour or so. You get a bit of red eye, maybe a little tickly cough. But when you start to get exposed to a higher concentrations it can get quite painful to breathe and breathing becomes much more difficult. When you go to a very high level - we’re talking about 300 parts per million or so - you can actually die within 30 minutes of exposure.

People often talk about antidotes to poisons but, I think, there’s only maybe about a dozen antidotes in the whole world for very specific chemicals. Unfortunately, chlorine is not one of them. So, basically, if someone gets exposed to chlorine and has these severe effects the only way to treat them is supportive. You give them some oxygen, you may want to give them some drugs to open up the airways, which are known as bronchodilators, possibly steroid to try and reduce the inflammation. And, if they’re lucky, it can take maybe three, four, five weeks to recover from a significant exposure to chlorine.

Music: Daniel Birch & Ben Pegley