IoT sensors, big data and advanced analytics to ensure optimal growth - Part II
In our second blog in the series “IoT sensors, big data and advanced analytics to ensure optimal growth”, we will read about how understanding human mobility can help us alleviate issues related to public health. The public health issue we will refer to is the use of illegal drugs.
Towards a better understanding of social issues
As we mentioned in our previous post, cities have been known to be the predominant engine and driver for innovation and wealth creation in a society. We mentioned as well that cities are also society’s main source of crime, pollution and health problems. To be better equipped to answer to questions like “How do we promote growth and urbanization while keeping crime, diseases and illegal drug consumption as low as possible?” we must have a better understanding of these problems.
Illegal drug usage as an epidemiology problem
According to the World Drug Report from 2012 (See United Nations Office on Drugs and Crime World Drug Report), about 250 million people (Nearly 5% of the world’s adult population!) are estimated to have used an illicit drug at least once in 2015. “Problem drug users” account for about 30 million, which is nearly 0.6% of the world adult population. By problem drug user, we mean that their drug use is harmful to the point that they may experience drug dependence and require treatment. At a global scale, illicit drug use seems to be generally stable (Even though it continues to rise in several developing countries). Illicit drugs undermine economic and social development and contribute greatly to crime, instability, insecurity and the spread of HIV.
Norway was the first country in Europe to establish a special clinic in 1961 for the treatment of drug users: the state clinic for drug addicts at Hov in Land (See Rusmidler i Norge 2016 for a report on the usage of alcohol, tobacco and narcotics in Norway). The target groups were the so-called classical drug addicts; healthcare professionals and patients with easy access to morphine and other opiates, that had developed abuse problems. In the mid-1960s it became clear that the use of drugs, and then primarily cannabis, was about to spread among young people. This new drug problem promptly called great concern, a concern that has not become milder in our present days.
Wastewater analysis and drugs
Wastewater analysis is a rapidly developing scientific discipline with the potential for monitoring real-time data on geographical and temporal trends in illicit drug use (See Wastewater analysis and drugs — a European multi-city study). Put in very simple terms, the method consists in sampling wastewaters to measure the metabolic residuals generated by the human body after consuming illegal drugs. It is a non-invasive technique that can capture the drug consumption habits of society, without suffering the biases and flaws of traditional methods, such as surveys. This method has been used to estimate illicit drug consumption in Oslo (See for example Sewage drug secrets revealed).
This method offers an interesting complementary data source for monitoring quantities of illicit drugs used at the population level, but it cannot provide information on prevalence and frequency of use, main classes of users and purity of the drugs. In other words, even if the method has great potential, it also has a few limitations.
Dynamic population size
A major limitation of the technique comes from the uncertainty in the estimation of the size of the population being tested. When considering Oslo, the way people commute to and from the city has a big effect on the size of the population present in the metropolis. The author of this blog has been involved in the research work developed by researchers at NIVA (Norsk Institutt for Vannforskning). In this work, we developed a technique to overcome the source of uncertainty related to estimating the effects of daily commuting patterns in the region of Oslo being served by the wastewater plant where samples to study drug consumption have been taken (See figure 1). More details can be found in the scientific paper Use of Mobile Device Data To Better Estimate Dynamic Population Size for Wastewater-Based Epidemiology.
Figure 1. Overview of the catchment area of the VEAS wastewater treatment plant. Figure taken from "Use of Mobile Device Data To Better Estimate Dynamic Population Size for Wastewater-Based Epidemiology"
The study was conducted during the summer of 2016. Using the cellular network as IoT sensors, we were able to capture the commuting patterns of the population living and working in Oslo, which is illustrated in figure 2. Things worth noticing in this figure are:
- The number of persons within the city is never the constant number provided by SSB (~600k persons in 2016)
- At a macro level, there is a downward tendency. This represents people leaving the city during Summer holidays
- The number of people leaving the city during the holidays was nearly 200k!
- We see a repetitive pattern: blocks of 7 peaks, with two of them being considerably lower. These are actually "working days" vs weekends
- In general, there are more people in Oslo during Sundays than Saturdays
- Once the city enters the holiday season, the dynamicity in the city is almost absent
Figure 2. Dynamic population of the Oslo (VEAS) wastewater catchment during June and July 2016. Figure taken from "Use of Mobile Device Data To Better Estimate Dynamic Population Size for Wastewater-Based Epidemiology"
What was learn
Figure 3 shows one of the main findings of our research. The line labeled as "Absolute loads -g/day" represents the total consumption of amphetamine, methamphetamine and MDMA by the population present in Oslo. When we add into the analysis the dynamic population size as a weighing factor, we obtained the line labeled as "Population Normalised Loads - mg/day/1000 people". Among the things that we learned from this we can mention the following:
- Consider the dynamics of the population a new pattern of drug consumption was discovered
- If one ignores the dynamics of the population, one can be drawn to the wrong conclusion that consumption was constant. We see that the consumption of drugs per capita increased during the holiday season
- If anti-drug campaigns had been run during summer 2016, the probability of reaching the troubled segment could have been much larger
- The author of this blog hypothesizes that a correlation between dynamicity and drug consumption that can be measured exists. Perhaps it would be beneficial for municipalities to organize more engaging activities that can bring people out of their houses, especially during the month of July.
Figure 3. Combined loads (absolute and population-normalized) of amphetamine, methamphetamine, and MDMA in Oslo wastewater. Figure taken from"Use of Mobile Device Data To Better Estimate Dynamic Population Size for Wastewater-Based Epidemiology"
We believe that the results of this paper represent an example where data science has been helpful to better understand a problem that is of great importance. Further experiments and research that take into consideration IoT sensor data and human mobility patterns should be developed to gain more knowledge and be better equipped to move towards solving the problem of illicit drugs.