#13: Japanese healthtech & AI possibilities to help people with dementia | Yu Shimizu, CTO at Aikomi

Published on August 31, 2022

People with dementia don’t become distant, they just lose their way of communicating. How can software possibly help them? Aikomi is a Japan-based startup that helps find a connection between a person with dementia and their family members via software. Yu Shimizu, CTO at Aikomi, tells about the challenges of an healthtech startup in Japan, its features, and the technologies that are used. She shows real-time cases of how AI helps create personalized content for people with dementia using a possible amount of data. Enjoy the listening!

In this episode, we will answer the following questions:

[01:15] how Aikomi works and how AI is used

[08:00] why at Aikomi don’t use the word «patient»

[08:55] differences from other similar apps

[14:16] current development focus of the company

[15:41] outsource team experience

[18:04] AI team management 

[20:32] generated unpredicted results from AI and how it recognizes Asian people

[25:08]  how the Japanese market is different

[31:08] in-house vs. outsource approach: pros and cons

[32:48] Aikomi scaling plans in Japan and abroad

[34:51] the most interesting projects from the Okinawa Institute of Science and Technology Yu worked on

[37:54] how to become a good-skilled engineering manager

[40:26] Rapid Fire Round (3 questions)



Aikomi: https://www.aikomi.co.jp/

Yu in Linkedin: https://www.linkedin.com/in/yu-shimizu-499149110/


Listen to the episode on iTunes, SoundCloud, Spotify, YouTube, Google Podcasts and let us know what you think on the topic.

The mission of our podcast is to show the real-life challenges of implementing technology in healthcare. And the podcast is sponsored by Demigos, a company that develops IT solutions for healthcare startups and companies. My name is Ivan Dunskiy, your host as always, and I am joined today by an honored guest, Yu Shimizu

Yu is from Japan, and her background is in applied mathematics, in biomedical signal and image processing. Yu spent most of her career in research, and currently, Yu is the CTO at Aikomi, a company that develops digital tools to improve the quality of life for people with dementia and their families. Yu, thank you for joining today. How are you?

Yu Shimizu: Thank you for inviting me. Thanks a lot. I'm fine.

Ivan Dunskiy: So could you please tell us about Aikomi and how it helps people with dementia?

[01:15] how Aikomi works and how AI is used

Yu Shimizu: Aikomi is a Japan-based startup. And we develop a web-based software. It's delivered by a tablet, it streams personalized content, and that content is specialized. It should trigger communication between the person with dementia and the family or the carer. And it in that way improves the quality of life for the families and also the carers, which is for us an important point so far. We want to focus also on the families and the carers, because many problems arise from this, what we call, dementia triad where families and person with dementia and care team, care workers have to work together to improve the situation. Essentially, with our software, we want to pinpoint the memories that are still alive in that person with dementia and show the family that they can still communicate with this person. And through that, find a connection back to their family members. 

Ivan Dunskiy: Basically memories. 

Yu Shimizu: Memories, but also cognitive functions. So the cognitive functions also decline, but, from outside, of course you cannot see which ones, still alive or which not, but you can trigger them, delivering person-centered content. So we also have quizzes for example, but with subjects that we know the person likes, we have an interview before starting the whole process with the family about the whole life story. And then, through integration of an AI, we try to eventually pinpoint what keeps that person motivated.

Ivan Dunskiy: Do you do that kind of in an empirical way? Where you show images or things like that and see, observe the reaction of a patient? 

Yu Shimizu: Exactly. Sorry, I didn't explain that properly, but the person who is watching the tablet is, recorded through the video camera on the tablet. And that video is processed by an AI to show what kind of reaction so far, not which kind of emotion they show because that's very difficult through an AI at the moment. We try to keep it objective, but in the future it will be connected to subjective comments, either from the family or from the carers who are with the person during the session. And in the next session, the AI should automatically select from a pool of contents that we already have. Something that's suitable for that person to continue that line of stimulus. So we actually call the videos stims like stimuli, we don't call them content or video or picture.

Ivan Dunskiy: Do you have evidence that these stimuli improved the cognitive functions of patients?

Yu Shimizu: It doesn't improve cognitive functions, but we can pinpoint what is still working.

We don't have evidence in the sense of evaluated data yet. We that's a work in progress. We have of course empirical data, and experience on that. That's what the whole started, the whole process of making a company.

We do talk a lot with the families also, but we also have a proper research study in preparation that we are doing in collaboration with care facility where we'll go through the whole process of recruiting, 30 to 50 people. And then evaluated before and after the sessions. After using them three months.

Usually people don't use the software every day or something. They use it every week. So it's a bit of a longer process.

Ivan Dunskiy: So they have kind of sessions where they use tablet or phone and see different images, maybe sounds, videos, right? That is how it works. 

Yu Shimizu: Exactly.

So, what we want to aim is that the family member is also interested and says: "That's interesting. I didn't know that about your life," or "I didn't know that you are interested in that." And that that's where the communication comes alive. And the second point is they cannot upload their own images. So if they send us their personal data, we uploaded and we make a story for them.

So they can look at events like 20, 30 years ago and share that memory. And kind of feel connected to that person anymore again. For the family, it's very difficult once the person starts having dementia to identify that person they knew, they mostly feel like they're losing that person, but it's actually not true. So characteristics are maintained. 

Ivan Dunskiy: What do you mean by characteristics are maintained? 

Yu Shimizu: If that person is inherently a positive person, because they have dementia, they don't become suddenly... 

Ivan Dunskiy: Negative.

Yu Shimizu: Boring. But to you, it might seem like that, because, they lose their way to communicate. So Aikomi should provide that bridge to show the family that they still can maintain their personalities. Like, people who make a lot of jokes. They still make jokes, even if they have dementia, that type of characteristics. People who are naturally quiet, they usually stay quiet. People are naturally active, they become active. But some of the symptoms of dementia, they kind of overlay that character like quiet people can be very aggressive, but out of frustration. So if you can kind of alleviate a little bit of anxiety and the frustration, then you can see their character come back. So that helps the family to kind of keep the quality of life.

Ivan Dunskiy: And, don't you think that if these sessions, would be more often, the result would be better? I mean, if that would be not once a week, but for example, every day. 

Yu Shimizu: So that is something we are not sure about, and we don't want to push the families to do that.

Ivan Dunskiy: I mean, not families, but patients. 

Yu Shimizu: There we are not sure we have to do a proper research project. That will be one of the parameters that we can't adjust. But from our experience, every day is usually too much burden on the care. So that that's a point that we find a little bit difficult, but which is a very important part of the whole business strategy also to find the sweet point for the caretaker and the family and the person with dementia.

[08:00] why at Aikomi don’t use the word «patient»

We don't use the word patient. It's always person with dementia. Because we believe that their symptoms come out of anxiety and frustration. And that's what we are trying to alleviate. We cannot fix the dementia itself. So we cannot revive functions that are already lost with this kind of software. 

Ivan Dunskiy: So a person with dementia cannot use the solution without a care taker, right? 

Yu Shimizu: If it's a very mild case still, they will be able. So we have a remote version, which we develop just for Covid, but where the family can actually be at home, or the person with dementia, if he live alone. They just have to switch on the tablet. The main aim is to make the family communicate. So we prefer if they are together. 

[08:55] differences from other similar apps

Ivan Dunskiy: Totally makes sense. And what differentiates Aikomi from other applications, which help people with dementia? 

Yu Shimizu: So that's a good question. That's the main point of the business also. So business of the product I would say is so we don't concentrate so much on the business. But one big difference is that it's person-centered so most of that application show general contents, general games, they're often group orientated for the care homes, but we try to customize it to each person and the customization works through AI. 

So many people ask us: "What's the difference to YouTube?" But the difference is that we have our own content of course, that's tailored for people with dementia and also that's tailored, to the personal content that the family gives us. So they can give us their own content and upload. And if they want also create their own stories, so we have more flexibility on the creation side for the family also. 

The second point is that we also want to concentrate on the quality of life for the families, cause many, many problems come out of the family, not from the person themselves. So the family would like to look after the personal, with dementia, but they don't know how. And, as you know, there is a shortage of therapists and also caretakers. So we want to provide the families with something they can use without having to do like a one-year training or something to have a good time with the family members. 

Ivan Dunskiy: Do you provide any kind of education or knowledge base to family members? 

Yu Shimizu: No. We on purpose want the software to be self-educational. 

So maybe there can be some kind of training on how to communicate more efficient, but I think that's very different from family to family. So we have family, for example, who communicate just by sitting closer. That's their way to communicate. We don't feel like, there should be any training. We are planning to have training for people who are working care homes because their relation to the person with dementia is a bit different than for family members. But it's still working progress also. So if they need some prompts, we are planning to put them into the program, not into a training session. So they will have maybe a sentence that the family can use to start a conversation instead of having to attend something before. We found that that's actually a burden for anybody who wants to try the software, if there is a training or if there is something complicated, then they tend not to use it. 

Ivan Dunskiy: Got it. I think that there is really a smart way to do it, because, as you said, the family members already have burden and have some kind of difficulty, right? So they're frustrated and that's cool that you give kind of plug and play way, approach where they can use the product right away. 

Yu Shimizu: It is actually also quite a challenge because our users are between 40 and 60, sometimes 70. So we try tailoring the software so that it's very intuitive. This is one of our challenges.

Ivan Dunskiy: And is the product already live? Do you already have users? 

Yu Shimizu: Yes, but it's in pre-business trial, so it's already live, but we only have like 30 users at the moment. So based on that, we will change the business model.

Ivan Dunskiy: 30 persons with dementia. 

Yu Shimizu: Yes. 30 paying customers. We have more people on the research side. 

Ivan Dunskiy: Got it. Cool. And what results do you receive so far? How do you measure outcomes? 

Yu Shimizu: We look at how long do people continue using it, and then, what kind of feedback we receive on the videos. 

We found something interesting that people tend not to continue using it if you don't check in with them. Which is one of our purposes. So the family feels more comfortable using the device. If they can call you once a week and say: "This session went well, this session didn't go well." Even though we don't advise them, but that's the type of service actually that we also would like to provide - a person who is familiar with problems in the family. If they have a person with dementia who can listen to them and not advise them, but be a communication partner for them. So it'll be integrated as part of our service. It's not a therapist, but it's somebody who is, well versed in that area. 

For example, our care director, she is the social worker who has a lot of experience with families like that. 

[14:16] current development focus of the company

Ivan Dunskiy: That's cool. Could you please tell us about the product development side of things? What do you do at the company right now on the product development side and what is your current focus?

Yu Shimizu: I'm in charge of connecting everything together and make it a functional thing on the web platform. So I tried to pinpoint what's needed on the business side what's needed on the user side, what's needed on the care side, and translate it into functions on the platform, but also into other applications if that's possible. So I'm also responsible for webpage. For example, we have a line for customers, anything that can be kind of solved by technology. I'm in charge of kind of coming up with that idea. Of course, I talk to my team with that. I'm responsible for communicating the needs of the projects to the tech team and figure out their workload, manage the roadmap, and manage the priorities. 

One of tasks that I nearly finished is to support the team to become a real team because we had an outsource team before. And we had to change to the in-house team, because we started with being only four people two years ago. We had to start from the first person in the tech team, and now we have six. So it took quite a while to kind of get everybody into the place they feel comfortable with, and feel also the whole process of developing. 

[15:41] outsource team experience

Ivan Dunskiy: Could you tell please about the experience of working with the outsource team? So, I assume that your technology is quite complex. I mean, there is of course, like some standard parts of it, but that is also that AI piece. So could you tell us about like the whole life cycle? What was the approach to develop the product from scratch and like how did it evolve to the current state? 

Yu Shimizu: We were quite lucky that the outsource team is actually a team that works for our collaborating company called SBX. The process was not like usual outsourcing where you have the whole spec and then you just give it to the outsourcing team, but it was like joint management. On my side, there was me, but they had their own manager, of course. I didn't have to know everything about what they're doing, but just tell them the needs. And then as the business changes also, of course in the beginning, tried to figure out how to do that. So they had the POC, the proof of concept, the base was outsourced. And then it was taken over slowly by our in-house developers. 

The AI is still outsourced to the same company and is connected through an API. And we continue collaborating in the sense of that we have weekly or biweekly meetings where we talk about how that process is and how it should be integrated in our side. So now it's completely separated into, the Aikomi team is in charge of the technology, technology in the sense of the software and the AI team is in charge of developing the AI and then connecting it through the API. So that now it's like two separate. There's like software development and one is technology in the sense of developing the algorithm. 

Ivan Dunskiy: So they do all the models. 

Yu Shimizu: They do the models and the data analysis.

Ivan Dunskiy: Got it. 

Yu Shimizu: That's where my expertise mainly comes in. So the data analysis and the machine learning, right. Because my background is not in software development. 

[18:04] AI team management 

Ivan Dunskiy: Could you tell us how you manage the team? When you manage the AI team, what kind of your approach to manage them? How do you overview if everything is going into right direction, like maybe some KPIs. How do you do that? 

Yu Shimizu: For the AI team, they have their own manager, of course, also because they're from a collaborating company. So they have their management also. But from our side, we are connected by Slack. We communicate about everything they need access to, if they need some API, then our team will take care of that. If we need them to configure something so that it actually can be integrated in our platform, then we communicate that mostly about in Slack. My role is more to keep the overview on what's missing, what's not missing. Because the developers are very focused on their own thing. It's not possible I think for them to at the same time look at the roadmap, and the AI has in the sense no hard deadline, but the technology side does. I am also in charge of balancing these two things. My role is more on the Aikomi tech team of course. 

Ivan Dunskiy: Got it.

Yu Shimizu: Even though I'm not a software developer.

Ivan Dunskiy: But again, how do you overview the results of the AI team, how do you check everything? 

Yu Shimizu: They give us a biweekly report and then they gave us also a knowledge transfer seminar.

Ivan Dunskiy: Seminar?

Yu Shimizu: We have knowledge transfer meetings, they explain detail how the system is built so that our technicians can take it over.

Ivan Dunskiy: Got it.

Yu Shimizu: So the delivery is maybe that biweekly. We have milestones, so we check on that, of course. But it depends also on what the data brings. So sometimes we don't get the results that we want and then we have to change things. But mainly for the outsource, we have biweekly meetings. 

Ivan Dunskiy: Sure.

Yu Shimizu: And I'm directly in touch with the manager on their side. So if anything is about changing in their team, then we communicate directly. 

[20:32] generated unpredicted results from AI and how it recognizes Asian people

Ivan Dunskiy: And there is a thought that AI is a kind of a black box and it can generate kind of unpredicted results. So could you share with us if you have already observed some unpredicted results from your AI?

Yu Shimizu: Unpredicted results, yes. So it's very heavily dependent on the training data, of course. And because we don't have enough data on our side yet, we use training data that's freely available also. So there is definitely a bias and I'm not even sure if I'm allowed to say that, but we trained, also previously, an AI that is supposed to tag what's in the picture. It's a very common problem. But we use it for Asian people, right. And the first time I uploaded an Asian person, it would say "bear". That's completely, and you can tell: "Okay, this must have been trained on white people probably." Right. These are kind of things we have to be careful about, I think. And also the training on emotional part. We try to streamline the training data a bit so that there won't be any mistakes like that.

Ivan Dunskiy: Just a similar example. We did a project where we analyzed the queue of people in front of post offices. We also built a face recognition algorithm to do that. And when we had tried Asian people, there were things like that, like you explained. It did good job with European people, but it was completely disaster with Asians.

Yu Shimizu: So, I think one big part of our project is also to customize the AIs to specific data because people with dementia, for example, they don't show the emotion that clear in the face. So even if we have a good emotion detector for face, but we have to retrain it on people with dementia. Because even if they say: "I'm very happy", you would maybe not be able to tell it. So that's where the main part of the AI goes into. 

Ivan Dunskiy: But, in general, what is your strategy of like solving this problem of getting more data? Because I think like more data you have, more accurate the algorithms, right? So what is your strategy in there? Is open data not enough? 

Yu Shimizu: It's not enough. So we use a lot of transfer learning. We try to find as much as open data and then adjust it to what we have. But we use of course also other data we already get from the clients. And then we ask friends too, to record videos. So it's still not enough, but we are not planning to have big data in that sense because it's per person. So there should be a constant update. For each person, we won't have that much data anyway. So we will have to find a different solution than a huge amount of data. 

Ivan Dunskiy: So you're saying that algorithm works only with the data of a specific person. It doesn't build correlations with other people.

Yu Shimizu: In the future. Yes. So we try to make it in part, it will build correlations once we have more customers.

Ivan Dunskiy: Do you think that is the right thing to build correlations? It is like two sides of the coin, right? This is the flip side that it can be biased. So you can think that it is a correlation, but in reality, there is no correlation. 

Yu Shimizu: That's very true. So I think, it would be a good starter for retraining the model. So essentially if there is a person we need to adjust in the future, now we are not doing it, but in the future, ideally there should be a feedback loop. If we can bring the start point as cost as possible to the end point, that's of course better for the model. So if there is a correlation, you can try to use that as a seed.

But there are many possibilities at the moment. We are not sure what's the best always but I think it's good to use as a starter. 

​​[25:08]  how the Japanese market is different

Ivan Dunskiy: Great. Could you please tell about the kind of your distribution strategies? As I understood, you are focusing on the Japanese markets right now? So what are your plans on going on other markets?

Yu Shimizu: So we need to establish the Japanese market first. We don't have enough people at the moment, because it's very culture dependent. Any content we stream is person and culture centered. So all the festivals are now Japanese related, even within Japan, because it's so long, we have different season, different vegetation, different festivals. So once we have that established, we want to of course, go abroad.

We already have a collaboration with care homeowners in the United States. And the feasibility project running. They will provide content and we will see what the first trial will show, how much different it has to be. And also the stimuli, they have to be presented in a different way probably. Especially if you compare Japan and America, the way we present things is completely different. 

Ivan Dunskiy: Could you elaborate on that? Like how different, in what way? 

Yu Shimizu: I think Japanese content is more subtle. One thing that came up is for example, the American care home, they have a similar application to Aikomi, but it's not person-centered but they have a lot of discussion topics. 

In Japan we don't discuss so many things. So if you would ask, what do you think about this? You will get less feedback than if you give them a choice, for example.

Our emotional reactions are a bit more contained. That would affect, for example, also the AI, the emotional, the face recognition and things. So these parts need to be adjusted, which is more work than most people think actually. 

So you will have to adjust whatever we stream and then you have to adjust the AI. 

Ivan Dunskiy: That is what you mentioned that, to put it in different words, so that has like taken into consideration the kind of closed culture. You don't really know if a person really means what they show. 

Yu Shimizu: Exactly. So we are trying to also keep a connection to Singapore because we have some connection there, that would be maybe easier because there are also many Asian people. And another thing is we try to find Asian people who live abroad as users to start with.

Ivan Dunskiy: By Asian you mean Chinese as well?

Yu Shimizu: Yes.

Ivan Dunskiy: And do you see similarities between how the product could be applied in Japan and in China? 

Yu Shimizu: Not in terms of content, but in terms of AI. 

Ivan Dunskiy: Not in terms of content, what do you mean?

Yu Shimizu: So in content, we will have to change everything to Chinese events, Chinese festivals. 

Ivan Dunskiy: Got it. 

Yu Shimizu: Everything. I think, content is the biggest part also, because of the copyright. It was not so easy to kind of gather all this content, you have to search for it. And then if it's perfect, it's sometimes it's copyright and you can't actually get it. 

Ivan Dunskiy: But as I understood, the majority of content is provided by persons with dementia, their families.

Yu Shimizu: Depends on how many images they have themselves. So a lot of content is also provided by us. So the base content should be provided by us because if there is a person who doesn't have any images or does not use smartphones or where you can easily get.

Ivan Dunskiy: Get the ground. 

Yu Shimizu: The target population is like 40 to 60, right? So, and at least in Japan, many of the customers we have are in the eighties and nineties, so they don't really have pictures of themselves. So we need to find them. It is kind of a quest also to find pictures for these people from somewhere from the past. 

Ivan Dunskiy: So basically they tell you a story, for example, "I was there." And then you put together the content. 

Yu Shimizu: Exactly. The first time before they use this, we have an interview with the family and they tell us the life story more or less of that person. And we try to find contents about that person. Ideally that should be automatic. It's semiautomatic at the moment. We use text to do that, but in the future, it should be completely automatic. 

Ivan Dunskiy: How it could be completely automatic? Will you parse or search for images online? 

Yu Shimizu: Yes, online. Ideally, it should all be in our package. It should not be online, but be on part of the Aikomi service. So you have all these contents ready at the time we deliver . 

Ivan Dunskiy: You mean that a person need to provide the content to you or what? How you can gather all the content automatically?

Yu Shimizu: We have a pool of contents already, so we still in the process. 

Ivan Dunskiy: So you just repurpose the content. 

Yu Shimizu: That's also a little bit to your question. If we can find correlations but we will try to find age if you see in the profile, same age, same hometown. Then we already know what kind of contents they will maybe like, so we start from there. And then the rest is feedback. 

[31:08] in-house vs. outsource approach: pros and cons

Ivan Dunskiy: Got it. And coming back to the kind of in-house versus outsource approach question. So what kind of pros and cons do you see, do you find in those two ways? And why did you end up with the approach of going more in towards in-house? 

Yu Shimizu: Outsource of course there's always a hard deadline and it's much more about budget related than in-house because in-house is continuous salary. And the communication of course works much better in-house. Even though we are all remote, because of Covid these days, it is much more easy to convey what is needed and for whatever reason, because they are familiar also with our business model. 

The in-house people, they know what the situation is regarding of what we aim to. While the outsource people, - we don't have as much time, we talk about the business also, or the needs that the care people have to make it more transparent to them, why we need certain functions. But we don't have much time. And the in-house people there inherently, they get this introduction to the whole company philosophy to our aims and our vision. It's more transparent for them. Their motivation is different. 

Ivan Dunskiy: You plan to continue development more in the in-house approach as a department role? 

Yu Shimizu: Yes. Except for the AI. I think everything will be in-house.

[32:48] Aikomi scaling plans in Japan and abroad

Ivan Dunskiy: Got it. Could you share with us the scaling plants of Aikomi? Do you have a roadmap, for example, when you go to the US market and what do you need to accomplish to get there? Just in general roadmap. 

Yu Shimizu: So we are now in series A. That means that until the end of fiscal year, 2023 we have to show that we have a business model that works. And then we go into series B, which is scaling. I think, we already have prepared the connections to different countries. But, realistically, we can only start, thinking about going abroad in maybe two - three years. 

Because we have to still grow our team and we have to still grow our marketing team. And somebody has to do the communication. We will need. If we go abroad, we will need translators. Maybe not so many, but it's still a bit further away, I think, going abroad. But we are counting on that actually. 

Ivan Dunskiy: Sure. I think there is kind of a global, as we all see, the global shortage of nurses and care takers. And I constantly speak with people from the US, and that is a huge problem there as well as society also getting older, in developed countries. I think that is a huge problem.

Yu Shimizu: I think in that sense, it's in Japan, the need is maybe much higher than in other countries yet. Japan is the first country where there's like a huge amount of old people and society is going that way because healthcare system is getting better and, people live longer. So I think everybody is watching Japan, how they're going to handle this problem. So it's a difficult place to start, but also a good place, I think, for a software like ours. Motivation is definitely there.

[34:51] the most interesting projects from the Okinawa Institute of Science and Technology Yu worked on

Ivan Dunskiy: You previously worked at the Okinawa Institute of Science and Technology, and you have some materials on the Research Gate platform. Could you please tell us about the most interesting project? If you can share that. 

Yu Shimizu: Right. I worked on two big projects. One is in pinpointing biomarkers in the brain for depression and the other one is completely different. I did both of them about two and a half years, but one of them is, the other one is in sustainable living architecture. And what I can say is that both of them were interested in the way that it took me very close to a real situation in the sense that, they are both topics that impact this society immediately at the moment in a very big way. And whatever we do as a researcher in that field can equally impact the society. Immediately. So that was my first experience in research. You do, of course something that's useful, but you're not often not so close to, even in applied mathematics, you're not so close to something.

Ivan Dunskiy: To implementation.

Yu Shimizu: Right. So it took me very close. I mean, the last step was done by going to Aikomi. I mean, still, and if you're in research, you can't see the last step where it gets implemented at some point. But you are very close. So I had the possibility to work with the therapists, the psychiatrists in the dementia in the depression projects. Also, on the sustainable living side, I had the opportunity to collaborate with all the companies that install solar panels. What kind of problems are there? Yes, wind energy. What is the problem? Okinawa? Why is not everybody using a solar panel? So all these things that in your daily life, you think like, that's important, but why is it not working? I got the chance to actually get the answer to that. Actually talking to people in on who directly work with this problem. So that was I think a big change also for me in the way I do research.

Ivan Dunskiy: And, I think that is great background to what you are doing now. 

Yu Shimizu: Exactly. So I used, especially in the sustainability project, that was the first time I got to manage.

The project myself. So that gave me the skills that I needed for Aikomi and also I had this collaborations with the companies that provide the hardware. So that was my first experience, because I'm a data scientist, I don't need to care about hardware. And I also got my first experience about software development there.

[37:54] how to become a good-skilled engineering manager

Ivan Dunskiy: We are coming to the end of the interview and, I would like to get your opinion on the question. What do you think it is important to become good-skilled engineering manager? What kind of advice you can give to people who want to switch, for example, from scientific jobs or from engineering jobs to engineering managers jobs?

Yu Shimizu: If the people want to switch, then they should try and do some management. Any kind of management, like organizing an event or just leading a group of kids in soccer, just to see what management is like, and maybe talk to other managers, what that job actually concerns. And then as a manager, I would say, the most important thing is to know your team. To know what they need from you and to know what motivates them to do, actually what they need to do. And if they can, if they are being able to do that they're themselves, then I think it's important to step back a little bit so they can have ownership, and control over their own process. I think too much management is maybe too much. This balance is very important. You have to know your people. It's not just about getting the things done. 

Ivan Dunskiy: But if you have a lot of people? 

Yu Shimizu: Then you have to have more managers. That's true. I mean, I have now the maximum, like seven or eight.

Ivan Dunskiy: Yes. 

Yu Shimizu: But that's true. So we already start having groups within. Because we are small, actually, we try to figure out who is comfortable doing what or who, even if they're maybe not the best in this one, if they want to challenge it, then I let them usually do it because that's important for their career development also. But I agree with you that once the company has like 100, 200 employees, everything is different of course, but there's no way, there's a different management style. 

Ivan Dunskiy: I appreciate that advice. I also agree on that, that there is not any silver bullet you can use, you need really to know, what is happening specifically with your people. There is no some kind of checklist that can follow and everything would be fine. 

[40:26] Rapid Fire Round (3 questions)

I'd like to end our interview asking several personal questions, if you don't mind. There is as just short questions and you can answer whatever you come up with. 

So I'm wondering, have you been to the restaurant? I don't know if I pronounce this right. Sukiyabashi Jiro and chef Jiro Ono in Tokyo. 

Yu Shimizu: No. 

Ivan Dunskiy: Have you heard about that?

Yu Shimizu: No. Sukiyabashi. 

Ivan Dunskiy: So I saw a movie, it's called "Jiro dreams of sushi." So that is about sushi chef Jiro Ono. That is the first sushi restaurant that received the Michelin stars, three Michelin stars.

Yu Shimizu: Right.

Ivan Dunskiy: And that is quite famous.

Yu Shimizu: Fantastic. I will check this next time. Let me see. 

Ivan Dunskiy: It's in Tokyo. If a restaurant has three stars, that is incentive to go to that specific country just to visit that restaurant.

Yu Shimizu: There is, you're right. You should come. 

Ivan Dunskiy: I'd like to. And could you please tell us what the location that impressed you the most in Japan is?

Yu Shimizu: I'm a very nature-bound person. That's why I'm in Okinawa. But the place I like the best is an island just below the main island of Japan. It's called Yakushima. It's now a national heritage. And do you know Miyazaki Hayao? He made the Princess Mononoke movie.

Ivan Dunskiy: No.

Yu Shimizu: Anyway, that place is really, really beautiful. There is a volcano, that's not active anymore. But when you start do the crossover, the scenery changes all the time. That's really beautiful. Everything is moss and green and you have the deer with the monkeys on top. So that's one scene in that movie, actually, of Miyazaki Hayao. There is a deer with a monkey. And when you go there, you can actually see them. 

Ivan Dunskiy: There is a famous anime, right? 

Yu Shimizu: I think you will know. It's called "Princess Mononoke." I'm sure you have heard. 

Ivan Dunskiy: I think that I saw other animes of him. 

Yu Shimizu: The forest that he pictures in these movies is always based on that forest. And everybody told me it's like paradise and I thought: "Ah, what's paradise?" but when you go there, it's really beautiful,the whole scenery always changed. And when you go to the top, you have this short bamboo grass, and then you can see over the whole sea. And when you go down, you have different more with water. And then you have the river, it changes all the time. It's really beautiful. 

Ivan Dunskiy: I assume there are many beautiful places in Japan. 

Yu Shimizu: Yes. 

Ivan Dunskiy: And what is the piece of advice you would give to your 20-year-old self? 

Yu Shimizu: That's a tricky one, but maybe to be more self-confident and give things a chance. I think I was not thinking too much by myself. You know, when you go to school, you just believe, at least me, it took me a long time to get out of this, everything that people tell you is true. Well, I realized later that's actually completely not true. What you learn is just should be the starter for your thoughts. But I took it as just that's my thoughts. I think, if I have another chance, I would like to be more open-minded from the beginning. 

Ivan Dunskiy: And I'm just curious, do you think that is also a part of the culture? Because I think that the Western culture is like more individualistic. 

Yu Shimizu: Maybe, but I grew up in Europe actually.

Ivan Dunskiy: In Europe. 

Yu Shimizu: I grew up in Vienna and I went to local school. 

So it is, but you're right. It is also culture related. In Japan, we are mostly educated to believe what your parents tell you. Which is also okay. That should be a nice in between. But considering that I grew up in Europe, I should be more open-minded I think. 

Ivan Dunskiy: Right. Thank you. I think that is the perfect way to end today's interview. Thank you, Yu, for your time and for sharing insights, of course about Aikomi and your approach of helping people with dementia, their families, as well as how you manage the product development, having two teams in-house and outsourcing teams. Thank you. But before we finish, what is the best way to get in touch with you? So that people can write you. 

Yu Shimizu: LinkedIn, I think. 

Ivan Dunskiy: That would also be in our sources section. Great.Thank you, Yu. Thank you, listeners. And we'll catch up in next episodes. 

Yu Shimizu: Thanks a lot. Thank you. See you.

Who is behind the HealthTech Beat podcast

We are a team of IT professionals who like sharing technical knowledge with healthcare industry people.

At Demigos, we generate ideas on how to improve product performance, design, and positioning based on our experience building complex health tech solutions.

Check our blog with articles on the related topics, and our cases in healthtech. Also, connect the podcast host and the CEO of Demigos Ivan Dunskiy on Linkedin.