Artificial intelligence or AI is a branch of computer science dealing with the simulation of intelligent behaviour in computers, or the capability of a machine to imitate intelligent human behaviour. It has been a very hot topic recently due to its capabilities, such as driving cars or being household assistants (Siri, Cortana, google home, etc.). In this essay, I am going to explore aspects AI in healthcare: how it has the potential to manage huge quantities of data and to help to deduce any latent diseases in patients.
A simple use of technology in healthcare nowadays is for data management: when nurses make an observation, instead of recording it on paper it is saved on an electronic database, individual for each patient. The data can be symptoms or medication which has been prescribed. Lab reports can also be stored on the database. For instance, blood samples can be analysed and saved in a section where every blood sample has its place and name. Continually, wearables have increased in popularity. They often come with heartbeat sensor, which collects data from the owner. This is another example of how AI can be utilized easily to improve health.
The problem that scientist are facing is that there is a great difficulty in having a huge amount of data stored and actually doing something with it; how to employ the data in order to enhance and make a difference in healthcare.
There are two ways of processing that data: a method called Bayesian nonparametric (for example weather forecasting, financial forecasting, etc.) this gained popularity five years ago, and Deep Learning, which is becoming very prevalent (for example self-driving cars, etc.). Deep Learning could be the solution to make the analytics of the data, but it is a long and painstaking process that we haven’t yet managed to initiate.
The ideal outcome would be to summarise the data in shapes to make them comparable. Machine learning is very good at that job, it can recognise patterns and quantify the shapes in people’s data. Thanks to that we can now model a stable human being. When new patterns emerge, we can compare them to see if they look like the old ones, so we know whether that person is stable or not. The solution would be to build lexicons, gigantic dictionaries of very stable patients categorised on age, sex, medical history, etc. This is what humans are not so good at, imagine doing for hundreds of patients; it is not within human capabilities or desires.
The situation in China is fascinating; it has an enormous population with huge hospitals, but thankfully they are very well resourced in terms of finance. The hospitals can make a swift switch to fully digitised electronic systems. This is not the case in the UK. Hospitals need to keep the electronic health record of their patients, and they can’t move away from them so easily. This situation opens the door to gather information from Chinese hospitals and transfer them back to the NHS so that everyone benefits from those results.
In conclusion, I think that healthcare services are making progress but that it is hard to obtain the outcomes mentioned. Creating a gigantic data base can be very advantageous, but it is difficult to make one, and it is a very time-consuming process. Collecting all the data for only one patient can be arduous as you have to take everything into consideration. You must indicate his sex, age, diet, life expectancy, previous contracted diseases, nationality, etc. When attempting to compare two patients, it can be tough because you must first consider if they have a similar diet, medical history etc or the number of variables that could influence your conclusion will deter your research useless. And because each patient is so different, the possibility of that patient showing symptoms unique to them is high. But, if in the future a digital database can be constructed which is capable of synthesising millions of patient’s data into a useful tool to help doctors with things such as diagnosis and prescription, lots of lives will be bettered.
lecture “AI in healthcare” by Prof David Clifton