[SOUND] When we talk to people outside the world of public health, particularly those from clinical backgrounds, they'll frequently say, public health, that's about data, right? While that's certainly true, we do a lot more than crunch numbers. I want to tell you about the field of public health intelligence, what it is, how it works, and why it underpins almost everything that we do in public health practice. For many public health professionals, health intelligence is an important part of what we do. For health intelligence analysts, it's a specialized career in itself. As you've seen so far terminology in public health varies between colleagues, let alone around the world. But I want to define health intelligence and harmonize the understanding of this sub specialty so that anyone who's trained here at Imperial knows exactly what health intelligence is. Now there's no agreed definition of health intelligence nationally or indeed internationally. And this leads to problems whether it's about resource implications, skill sets, or job titles. But for me, health intelligence is about generating actionable information. Let's take that apart. This is a diagram for my colleague Chris Williamson. He's amassed many years of experience in this field. I don't need you to learn this diagram, but I do need you to understand it. The world today has more data than it can ever possibly even process, let alone understand. It might have been your local transport system recording your journey. It might have been you uploading a photograph to your social media account. It might have been a multinational company recording how long you spent looking at a web page, that new item you wanted for your birthday. The world is collecting data every second of the day. With evolving data protection legislation in Europe, you can now require a company to give you all the data they hold on you. A few months back, I asked a well-known social media platform to give me all the data they held on me. Among the data they sent me was a long list of IP addresses. These are strings of numbers with what look like times and dates. Now, if you don't know anything about IP addresses this sequence of numbers and nowadays letters is entirely meaningless. But if we can find someone who knows what an IP address is and how it can be used to geographically identify someone, suddenly these data take on some level of meaning. If we then begin to add in the date and the time stamps, we can begin to develop a picture of where I was and when based on my log into this social media platform. Some of you will be saying well that's obvious. But what I want you to understand is that data of themselves are of little inherent value and that's where we come in as public health specialists. We add value by interpretation and then by answering that question, so what? Let's look again at the intelligent cycle. We start off with data. In this case, let's take the admissions data for patients presenting with heart attacks, what we term medically myocardial infarction. The admission data provide for a patient number, gender, age, time, date, and length of hospital stay. So how is this turned into information? Well, you have to understand what the variables actually mean. For example is the age variable based on age now or age at admission? Are the times recorded using 12 or 24 hour clock? What do blank cells mean? Has the data set been cleaned and coded? This might seem really obvious, but believe me as soon as you enter real world public health, you'll be emailing the person who sent you the data set to ask precisely these questions. So how is this magically turned from information into intelligence? Well, it's about understanding context. The information at the moment tells us how many admissions there have been, but is 150 good? Is 150,000 bad. Well, we have no idea. We have to add context and manipulate these data. You might want to add a denominator, overpopulation and time period with these data collected. You might want to generate incident statistics. You might want to compare your group with another similar geography or break down your group into subgroups based on gender or age. It's only now that you're turning the information into something that's actually beginning to mean something, but how do you know what to do? How do you know what you're looking for? Well, this is where your public health training comes in. You may know that myocardial infarction typically occurs in later life and is more common among men than women. By introducing that knowledge, you can begin to generate hypotheses or lines of inquiry that enabled you to interrogate the data. This brings us to the next step about policy and answering that question I posed before. So what? This is by some way the most difficult part of the whole cycle. I've worked with many people who are new to public health, but are brilliant numbers. They can handle a spreadsheet very capably, and they bring me statistics and numbers and brilliant analysis. I want to quickly think about that word, analysis. Analysis comes from the ancient Greek words meaning to break things down. And while it might seem difficult to you right now to even do the analysis correctly, that's actually not very difficult. You'll be able to give me the incidence of myocardial infarction, men versus women, under 45 years and over 45 years in age. The opposite of analysis and also from ancient Greek is synthesis, to place things together, more commonly to build something up. The process in health intelligence is to analyze problems and synthesize recommendations, or even better, generate solutions. Let's go back to this idea of so what. Well, if the incidence in your population of myocardial infarction is higher among women than men in contrast to the normal position then what can you infer? My first instinct would be to understand precisely what incident statistic we're using. Has it been standardized by age for example? I'd want to carefully understand precisely what the inclusion criteria were for the numerator and for the denominator. Once I understand the situation I can begin to suggest what's happening. Is it to do with coding? Is it to do with demography? Or is it to do with case makes? Once I've developed my picture including my theories of what's going on, I may want more expertise. I might call a colleague working in the community or a cardiologist in the hospital. You see it's only now that we can begin to figure out what to do about it. The answer to so what is because of X, let's do Y. Now the rest of the cycle is straightforward. You enact changes, and you continue to monitor the situation starting the cycle all over again. Health intelligence is inherently quantitative. But to be good at numbers is necessary, but insufficient. Health intelligence specialists need to have a balanced set of skills, data manipulation, data analysis, information governance knowledge, public health interpretation skills, and most importantly, the ability to engage, communicate, and collaborate with a wide range of professionals to effect change. Now that skill set is pretty rare. Health intelligence is applied epidemiology. It is crunching numbers, but it is so much more. [SOUND]
