To contribute to society by establishing a method to measure pain
with universal indicators
in order to provide safe and comfortable medical care equally to all patients.
Pain is one of the most unpleasant sensations and serves as an alarm to warn living organisms of a crisis.
In order to properly evaluate it in a timely manner, it is necessary to receive proper medical care by assessing the condition accurately.
On the other hand, long-lasting pain is not an alarm.But for a patient, it could be very painful, and often includes suffering or agony because it is misunderstood by others.
Since prolonged pain is harmful to the living body, appropriate treatment is vital.
As of now, there is no method that can evaluate pain objectively; it is basically defined by self-declaration.
If a person says it hurts, or it is painful; then there is pain.
The sensitivity and expression of pain vary greatly depending on individuals, so patients sometimes fail to receive appropriate treatment.
Other times, inadequate pain treatment causes severe side effects such as being in a vegetative state or drug addiction.
Since the process of pain recognition is highly complex and not easily clarified, elucidation of pain recognition may have to wait for a much farther future.
一On the bright side, great progress has been made in extracting a brainwave feature.
with the power of Artificial Intelligence, we are developing a system which can evaluate pain
without taking verbal information from the patients themselves by picking out pain components from large set of pain data, including its intensity or severity.
What we can do to achieve the world standard for pain treatment
Current method of pain evaluation
Pain is one of the most uncomfortable sensations that a person feels, but there is no way to measure people around the world with the same measure as you can with vision or hearing tests.
The only way to judge a patient’s pain is to ask the patient “How bad is the pain?”
Because we rely on the words patients say, patients who are hypersensitive to pain and patients who can endure a lot can have different expressions even if they have the same amount of pain.Thus, people who really need treatment can be missed.
Determining how much painkillers to use is very difficult.
If you make a mistake in this amount, it can lead to problems such as those shown in the table below and can even be fatal.
Three problemscaused by lack of “Objective Pain Judgement Method” due to reliance on patients’ self-declaration
◆Overdosage of analgesic
Pain monitoring system + Proper dosage → Avoidance of serious accidents
◆Underdosage of analgesic
Pain monitoring system + Proper dosage → Avoidance of complications
◆ Analgesic Addiction
Pain monitoring system + Proper dosage → Avoidance of addiction
In addition, patients who are particularly tolerant to pain may not be treated properly, or serious illness may be overlooked.
The services that are really needed are not delivered fairly to those who really need them, and this leads to the disadvantage to society that we cannot save when we have to save.
Pain is felt in the brain. Therefore, we have been working on the development of a device that “visualizes” pain using brain waves.
We think what we are developing can be called an “automatic pain detection system” because it has a function that tells us if it is painful or not without saying anything. This is the world’s first automatic pain detection system using EEG.
Automatic pain discrimination system
Module 1：Automatic pain discrimination system for hyperacute phase
◆ Treatment under local anesthesia ◆ Dental anesthesia ◆ Treatment during endoscopic surgery
Confirmed effectiveness in clinical research
Module 2：Automatic pain discrimination system for acute phase
◆ Determination of pain after a surgery ◆ valuation of drug efficacy (e.g. clinical trials) ◆ Bedside use
Preparing for doctor-led clinical trials
Module 3：Automatic pain discrimination System for chronic phase
◆ Detection of long-lasting pain ◆ Evaluation of drug efficacy （such as clinical trial）◆ Use in outpatient services
Embark on system development
If it becomes possible to properly evaluate pain by using this automatic pain discrimination system, achieving the world standard for pain treatment (establishing a method for treating the same pain for people all over the world) is not a dream. Doing so will improve the quality of life (QOL) of pain suffering patients around the world.
That is the social contribution that PaMeLa Co., Ltd. is aiming for.
Achievements to date
We have been aiming to visualize pain in a way that is easy for anyone to use, and does not cause burden to a patient. We have succeeded in visualizing pain by developing a tool that objectively evaluates pain using brain waves (EEG), which is one of the technologies used to capture the signals emitted from the brain.
■ 1) Hyperacute phase pain evaluation
For small surgeries performed while a patient is awake, the patient first gets an injection of anesthesia. The injection hurts because it uses a needle. During the injection (local anesthesia), we asked the patient to use “Visual Analogue Scale (VAS)” to evaluate the amount of pain felt by the patient with a numerical value from 0 (no pain) to 100 (maximum imaginable
pain). Then, the EEG electrodes were applied and the brain signals were measured. The pain that the patient felt during injection was evaluated using a method called LSTM (Long short-term memory), a type of deep learning in artificial intelligence. And it was found that the pain component can be extracted from the brain signal when pain occurs. This achievement was selected as an excellent presentation at the 53rd Japan Pain Clinic Society in 2019.
■ 2）Evaluation of pain at “high level”
Using machine learning within artificial intelligence, we determined whether four different magnitude pain stimuli in 146 subjects could be distinguished from the most intense pain. The subjects with a relatively good fit showed about 80% of the almost complete answers, and the overall discrimination accuracy was 67%. In low-fitting models, the problem of noise that comes out naturally from the body was found. We are currently working to make a better distinction.
■ 3）Discrimination of “qualitative change” of pain stimulus (discrimination of unpleasant pain)
A person may not need a treatment just because pain is intense. For example, a massage with a strong pressure may cause some pain yet also feel good, but such pain does not need to be treated.
We gave pain stimuli of the same magnitude but different degrees of unpleasantness (unpleasant yet feel-good pain, not detrimental to body) to distinguish between unpleasant pain and pleasant pain. For this distinction, we used a method called support vector machine, a kind of machine learning algorithm (calculation method programmed by computer). First, a determination model was created using the data (learning data) of multiple participants, and the unpleasant level (2 levels) of the new test data (unused data in the learning data among the data of the people who want to be distinguished) was estimated and discriminated using the feature quantity represented by alpha waves, a frequency in the EEG signal. At the current stage, about 70% can be distinguished between unpleasant pain and pleasant pain. This way, even if the pain level is the same, we can pick out the pain components of those who need treatment.
■ 4）Verification of pain reduction effect through β-endorphin by exercise task
In order to check whether the equipment we develop works well, it is an important to know that pain relief could be measured or not. However, the data we received from patients showed different degrees of pain felt by them. Therefore, it is not suitable for development using artificial intelligence. In addition, from an ethical point of view, it is not acceptable for healthy participants to take analgesics and collect data. Hence, it was necessary to reproduce the same situation as when conducting a drug administration experiment in order to develop a highly accurate device. For that purpose, we tried to use strenuous exercise as a mean to verify the analgesic effect by promoting the natural secretion of endorphin, which is known to be secreted in the state of runners high. Through this study, we were able to verify that using our technology under development, the pain score could be calculated based on the EEG.
■ 5）Attempt to objectively evaluate chronic pain
With the cooperation of patients, we have confirmed in a preliminary study that this method using machine learning can be applied to the objective evaluation of chronic pain.
Current development status
We have developed various signal processing techniques to make pain visible. Four patents have already been granted.
Pain using machine learning (sparse model: Japanese Patent Application No. 2017-137723, support vector, neural network, deep learning: Japanese Patent Application No. 2017-254565) and new acoustic signal processing technology (Signal enhancement technology: Japanese Patent Application No. 2017-148350) We have developed a method that can automatically discriminate the intensity of pain, and whether the pain is unpleasant or not; and also have developed a new method of extracting pain (electroencephalogram feature amount) from the electroencephalogram that can make the discrimination more accurately. Using this method, it is possible to visualize the pain as a numerical value while one begins to feel and continues to feel pain.
Since such a system will lead to more accurate evaluation as the amount of data increases, for the purpose of creating a mechanism to collect data of patients using our developed equipment and to turn it into big data, we have applied for the patent 2018-2777.
We have developed a model using a relatively weak reference stimulus (light pain stimulus given from the outside) to improve accuracy, and developed a method for calibrating (adjusting to match the standard) the feature amount of the signal extracted from the electroencephalogram. For this technology, we have applied for Japanese Patent Application 2019-85779.
Furthermore, we have developed a data amplification method that enables machine learning with a small amount of data, and applied for Japanese Patent Application 2019-85782.
With the aim of developing a device that can be used in a hospital and in a clinic to measure and visualize pain,we are working on the ease of use so that the device is easy to put on and take off, while learning what kind of devices are widely adapted in medical field. We are also working on the development of compact devices that are less burdensome to patients, about the size of a small electroencephalograph.
Flow of an experiment (example)
Experiments to measure brain waves are conducted in the following way.
We use special equipment and pay close attention to safety so that participants can take part in our experiments with peace of minds.