Monday, September 29, 2008

MIT NextLab Video: Solving medical imaging workflow challenges in rural Zambia

Here is a video that illustrates the challenge we're working on with CIDRZ in Zambia. Great job by Nicole Prowell and Elliot Higger, our Emerson communications liasons on the project in the production of this video in framing the problem and presenting the team we have available.

Team Mobile Diagnostics Project 1 from Nicole Prowell on Vimeo.

I've previously posted on the solution we're coming with in these posts:

Workflow management for medical imaging in the developing world

Open innovation for developing world health care

And here's the website if you want to keep track of our project:

NextLab CIDRZ Mobile Diagnostics for Cervical Cancer in Zambia

Saturday, September 27, 2008

Analyzing the value of pricebots

Here's an exercise that I did for Economics of Information at MIT Sloan that shows the value of pricebots. The item that I was searching for was an Apple iPod Touch 8 GB (2nd Generation) including shipping (assumed). The key insight coming out of this is that pricebots can provide a user with significant value for their time. The one I especially liked coming out of this that I didn't know about before was MySimon, which I found very intuitive and easy to use.

First, I searched a few random sites for prices: $219.52 58 seconds
eBay $189.95 65 seconds $229.99 23.8 seconds

Amazon: It was confusing between the 1st and 2nd generation 8 GB iPod touch. Actually, I came back later and found $210 + $4.99 shipping from a just launched Amazon Stores vendor. Regular Amazon unit ships in 8 to 9 days.
eBay: In the initial view, it is hard to separate between the Auctions and “Buy it now”. I chose to look at the Buy It Now prices rather than bid through any auctions. I’m pretty sure this is a 2nd generation one that I found at $189.95 that includes shipping but not completely sure.
BestBuy: Very fast and intuitive search, but the highest price.


I just took a look at two pricebots and ran the quick analysis for price dispersion and the "wage" given for using the engine.

Best Price: $219.95 from PC Connection 2 minutes and 4 seconds
I was a bit confused by the interface and ended up looking at used/refurbished units before going back to find the link to the new units.


Circuit City: $229.99
PC Connection: $219.95
Best Buy: $229.99
PowerMax: $228.88
Vann’s: $229
iUnitek: $229
Abt: $229
Time required: 44 seconds

$219.95 from PC Connection (best in stock price) $219.94 (Listed as out of stock – actually 8 to 9 days)
Apple $229.00
Circuit City: $229.99
Best Buy: $229.99
Abt: $229
Vann’s: $229.00
iUnitek: $229 (Out of Stock)

Interesting comment: Amazon’s prices show up as $219.94 on both these sites, but $219.52 on their site. Not sure what accounts for the minor difference.

At least with this example of the 8GB 2nd generation iPod touch, MySimon and PriceGrabber didn’t come up with radically different prices.

Absolute Dispersion: MySimon: Highest price was $229.99, lowest, $219.94 if you are willing to wait a few days. $229.99 - $219.94 = $10.05. On PriceGrabber, it was $10 - $229.99 - $219.99.

Relative Price Dispersion: Price dispersion is 4.3% for PriceGrabber and 4% for MySimon.
Mean Market Price: $227.98 for PriceGrabber, $226.98 for

Shopping Wage per Hour

PriceGrabber: ($227.98 - $219.99) / 124 seconds = $7.99 / 124 seconds = 6.4 cents per second
MySimon: ($226.98 - $219.99) / 44 seconds = $6.99 / 44 seconds = 15.9 cents per second

3600 seconds in an hour so the hourly wage of using these sites as opposed to selecting a random vendor:

PriceGrabber: 6.4 cents * 3600 = $230.40 15.9 cents * 3600 = $572.40

Quite striking if you think about it this way. If I hadn’t gotten slightly lost on the PriceGrabber initially, the numbers probably would have come out similar in terms of the shopping wage per hour. My first choice for electronics is often Best Buy, so if I remembered to use a search based pricing engine, I might save a few dollars. Note that eBay didn't come up in either engine.

Analysis and Commentary

The iPod touch example is a bit of a generic one since it is such a mainstream product. I can imagine that if I were searching something odder, the value might even be higher as the natural price dispersion might be higher. For instance, some of the designer furniture that I often find myself hunting for in a cherry finish to match the rest of the items in my apartment.

This has pretty interesting implications for competition. We know that the web has put the power in the hands of the consumer, but the interesting thing is how to take advantage of it. Interesting that Amazon’s price is cheaper than all the others by a nickel on the search engines (though due to the 8 to 9 days to ship, it looks like they treat that price as out of stock and drops it to the bottom). This seems pretty genius when you actually think about it. I don’t doubt that Amazon has a pricing engine that looks at the prices on the web and undercuts others, assuming a set margin can be maintained. Combined with their operational excellence, this should allow them to capture a pretty substantial portion of the market without going any lower than they have to. The problem is that if other vendors do this, the price will be driven down to either marginal cost or the minimum acceptable price for each vendor.

Also, it seems to me that you would want to be able to charge people who come directly to your site more since it may be implied they are shopping on brand or for convenience. This may mean offering coupons to lower the price when people are using pricebots or some other way to differentiate between these two types of customers that would be acceptable in the minds of customers.

Of course, price isn’t the only factor. Quality of service, speed of delivery and ability to return items also matter to me. Sometimes using these pricebots, especially when searching for a less mainstream item, some of the best prices come from smaller Internet retailers. This is especially true with eBay – quality of service varies greatly, and I’d rather pay a few extra dollars knowing I can avoid a lot of hassle if there is an issue with the order. The $189.95 option leaves a lot of questions unanswered. I suppose I could e-mail the seller and ask, but that’s sort of a pain. That makes branding and track record more important. It can break ties or make the difference with price dispersion is low as with this example. For instance, if Vann’s at $229 and Best Buy at $229.99 were the lowest prices, I’d probably still buy from Best Buy as I haven’t heard of Vann’s. The fact that Vann’s has a number of good reviews and a high customer satisfaction on the site helps. Especially when getting into lesser known retailers, the reviews that pricebots have compiled on various vendors are helpful.

I want to throw one more pricebot that I use: Upromise. I probably use it more than any of the others out there. Upromise has a decent pricebot, but also provides me with a kickback into a “college savings fund”, ranging from 2% to 10% of the purchase price. Over five years, I have saved about $1,000 in my Upromise account. For some reason, this type of value add for me is more of a hook than the “hourly wage” way of thinking about it. However, I went on Upromise and searched for the 8GB touch and was thoroughly confused by which generation I was getting. So maybe I should forget the couple dollar kickbacks and use MySimon, which struck me as intuitive, comprehensive and well-designed. Maybe UPromise should just buy software from one of the other guys out there and integrate it.

Wednesday, September 24, 2008

Workflow management for incoming medical images in developmental health

Some updates on the MIT NextLab/global health project that I’m working on. Originally, CIDRZ and GE were supposed to be stakeholders, but due to bandwidth concerns, we’re primarily focused on the CIDRZ project and improving the workflow to diagnose and treat cervical cancer. However, having GE involved made us think about some of the more general issues that are out there, and we’re trying to build a solution that is scalable and solves a few of these.

One is that there doesn’t seem to be a good system out there that can manage image/video based workflow and take multiple inputs. As such, we’re moving towards a system that integrates workflow and the tagging, commenting and return diagnosis features. Think a Flickr like site integrated with records management that can accept incoming images a number of ways including e-mail, via mobile phone, web or USB upload. We'll try to build it so it can integrate relatively easy with existing medical record systems. Comparing it to Flickr is a bit of an oversimplification, since the application should allow for a lot of features, such as creating a queue for radiology experts in different locations to review, return a diagnosis, store medical records for future reference, etc.

We’re also exploring the possibly of using Eye-Fi or other technologies to accelerate the transfer speed and more importantly, streamline the steps necessary to transfer the data. One of the problems with existing smartphones is that they ask permission to run applications and send data. So applications that sling data back in forth in an automated manner help the process flow a lot.

We're realizing that this has the potential to scale out not just the CIDRZ model, but be re-usable for other rural health models. It's interesting to think about this in terms of being able to help millions of people. The Emerson communications team that is working with us to document the project will be posting some videos soon for anyone who is interested in learning more about the project. Any ideas or possibilities for collaboration are welcome!

Friday, September 19, 2008

Amazing photo collection of Hurricane Ike's damage

Check out this amazing photo collection of the damage caused by Hurricane Ike. Totally wish more websites would publish this quality of photo in their new coverage. Just remarkable. I had no idea the extent of some of the damage. You usually just see dollar numbers. These photos really put things in perspective.

Paul Kagame on technology, self-reliance and Africa's victim mentality

Paul Kagame, the President of Rwanda, spoke yesterday at MIT and I had the privilege to be there. It was quite an interesting talk - not typically the one you would expect from a head of state. Sure, he had to talk about Rwanda's process and improving economy, but Kagame was also quite candid about Africa's need for better leadership and increased self-reliance. As Rwanda's first democratically elected president, his words ring true.

The core of Kagame's talk was about how mobile technologies are transforming Rwanda's economy. Kagame points out that 28 million subscribers to mobile network have create an almost instant infrastructure for his country's entrepreneurial people. Business models being transformed. One example Kagame gave was fascinating - mobile payments for metered energy. Electrogas, one of Rwanda's biggest utilities as a power supplier, was once a money-losing, inefficient company that could not collect its bills. Now it sells metered energy through mobile payments at retail outlets, and is profitable. That means the money the nation of Rwanda was spending to subsidize the utility can now go towards more important goals.

According to Kagame, education is highest on that list. Rwanda is a participant in the One Laptop Per Child program, and Kagame believe that education and understanding technology is the key to bringing Rwanda and Africa in line with the rest of the world in terms of living standard. Kagame stated that Rwanda's goal is to be a middle income country by 2020, an ambitious goal, but a good one to set.

Answering questions from students (how many heads of state do that?), Kagame took quite a strong stand on Africa's mentality as a victim. He said, "Africa must stand on its own." He said that African countries must become stable and economically successful, and provide their own services, not just hold its hand out to the West as they have done for so many years. He said "It is as if we have become comfortable with that, that people will come from the outside to address our problem." He laughed at the suggestion that the Chinese are different from past trading partners, and certainly no worse than past colonial raiders. The implication is that it doesn't matter who comes to Africa for its resources, Africa must learn to take the revenue and transform its economies into self-supporting entities that can meet the needs of its people.

Kagame closed out by saying “Africa needs to stand up and make itself relevant.” That statement drew a standing ovation from the crowd.

Thursday, September 18, 2008

First Minute Clinic in Massachusetts

CVS and its MinuteClinic retail franchise cracked a tough nut in opening up Massachusetts to the innovative new model of retail health clinics. It sounds like they're in-network with many of the major providers, so they've gotten insurers on-board, which is key. Hopefully, this model can scale up and help take some of the costs out of health insurance in the US.

Wednesday, September 17, 2008

Superconvergence: thoughts on the digital music and phone businesses

Sometimes it seems like deja vu it's so common, but every business school class has a case on digital music and iTunes these day. Since I've written a lot on it recently, I thought I would try to compile my thoughts into a somewhat coherent post on my view of where the industry is going. These are just jotted down thoughts over the past week, but they may stimulate something thinking.

What's the future look like?

As a futurist, I imagine a converged device where through a phone or Wi-Fi connection (quite likely it’s looking, an iPhone) on a 4G network where any piece of digital content can be streamed to a phone. This is the subscription service that many would pay for and it could in the future be offered by a combination of a cellular network provider and an iTunes subscription service. Until then, I don’t really foresee a subscription service coming from Apple. If it happens, it will be because someone with the power to make it happen is able to offer such a compelling subscription service that users are willing to pay.

Content ownership shouldn't be an isue - DRM will die, and people will own their content and have plenty of storage. With storage increasing the existing rates, we'll have 500 GB mobile devices in 8 year or 10 years. That's no problem for music where songs are 4 MB files. Video might end up being a little bit more of an issue.

The more interesting question is whether Apple will get to control this market as well. There will be more heated competition in this space. However, handsets are more complex devices with many fierce and innovative competitors. In terms of a platform, they are much more like PCs with many different needs and demand for software. Apple’s closed platform strategy may face resistance as Google Android emerges, and other manufacturers develop innovative converged phone devices that meet customer needs.

What about existing subscription services?

David Pakman came in and spoke the other day at MIT about his company eMusic. eMusic is one of the smartest models for selling music out there, and I think they think about it the right way. Their customers spend on average $170 a year, and the amazing thing is that they don't have any of the big 4 record labels. This is quite amazing. What eMusic does have is record labels 5 through 50,000, a reasonable model to pay them, while retaining a reasonable share of the profits. They also have good technology to help adult users find music. It's great if you're a fan of jazz, blues, classical or any other Top 40 type music.

I'm not sure that in its present form this can be a dominant market player, but it's a smart way to build a business. Pricing wise eMusic has a subscription service with different tiers that give you a certain number of download. Pakman says these were figured out through research and optimization models. It's positive for the music industry because unlike what Apple does, it's a genuine effort to maximize revenue. That helps long-tail artists and record labels, which means more good music.

I still don't think subscription services will become wildly popular until they sit on a platform like the one described earlier in this blog post. That is, the integration with the platform (software and hardware) has to provide a compelling customer experience.

Will digital music cost a song for ever $.99 forever?

The cost of digital songs have been at $.99 for some time now. Given inflation, that means the real cost of music has been falling since this model has fallen into place, and indicator of the ferocity of competition and indicative of how no one has really answered the question of how to make money selling digital music (as opposed to selling music players).

The reality is that people are incredibly price sensitive in digital music market (hence all the people willing to brave the terrors of the Internet to download free music). Professor Richard Schmalensee, the former dean of MIT Sloan who is teaching Industrial Economics this semester mentioned -3 as the elasticity of price. For marketing and economics reasons, the price of a song has been set at $.99. If Apple even raised prices 1 cent, a large drop in demand or groundswell of disillusionment could be envisioned. The worst thing that Apple could do is hurt highly profitable iPod sales by making people who are locked into their platform unhappy. Apple should seek to keep market conditions as they are in the digital music and player business as long as possible.

But how about someone else? With the value of the dollar constantly deflating and expenses rising, how can the music industry not afford to raise their prices? The Big 4 record labels will eventually want to increase their share.

On the other hand, it's possible that the price of some music could go down. Some songs just aren't worth a buck (one could argue that people's WTP for many songs they pirate is zero). Some songs would be best priced at $.50. More bundling, either in the form of subscription services or selling additional incremental units of music make sense. Eventually, firms should be more innovative about increasing the amount of revenue (and hence profitability, since the marginal cost of selling additional music is virtually nothing except paying the record label, a little bandwidth and payment processing).

Apple hasn't been very proactive about this, but I think it's their best interest to do so. If Apple could figure out different ways to bundle songs to increase sales and albums, that ought to benefit everyone. Some ideas include dynamic bundle pricing or selling a higher number of downloads at a higher price. Part of the problem is that albums tend to have pretty bad reputations, a grave that the record industry helped dig themselves in their heyday. Myself and many other music fans felt that an album where there were more than two or three tracks worth buying was rare. When a CD was $15 and singles were $5, it was still worth it to buy the album just to be able to play it continuously without changing the CD. It’s different with the way digital music is played. You only buy the album if you believe that 9 out of 12 songs are worth buying individually, which rarely feels like the case. It might be better to accept the reality and get people to buy the 3rd and 4th songs off those albums (e.g. 99 cents for the first two songs, 50 cents for the 3rd, etc.)

What's going on in the existing market?

The state of the existing market is an interesting one. Apple’s dominant position as the platform provider means they own approximately 70% of the digital songs and music player market. While normally, these would be considered separate markets, Apple’s integration means that they are a platform with a significant advantage against competitors who generally only make hardware or sell songs. Even though some of these firms have tried to partner up to form platforms, they have been able to wrest much market share from Apple. This allows Apple to sell music at a small gain and increase lock-in to their platform, since Apple’s DRM means users would have to go through a major hassle to convert their music into other playable formats. Meanwhile, Apple’s control and popularity gives them tremendous market power in selling their innovative hardware.

What are paths to success in digital music?

Economies of Scale: The music game has historically been a game of giants due to the economies of scale required in distribution. In order to justify paying an artist and producer big money, a massive reach is necessary to maximize profits. Digital content allows for many of those intermediaries to be cut out.

Loyalty/Switching Costs: Apple’s strategy of creating a locked-in platform on both the hardware and software fronts didn’t work that well in the fight against Windows. This type of strategy makes it more difficult for open and outside innovation to meet a diverse set of needs (in the case of PCs, the software the drives businesses globally). However, for most people, music is a simpler proposition. An intuitive, elegant, integrated solution that is easy to use right of the box is ideal. It’s interesting that Apple continued to follow this integrated approach to win the music business. For RealNetworks, this means partnerships with hardware makers and operating systems will be important. This ended up being a major barrier for RealNetworks. They ended up in a messy anti-trust suit with Microsoft that was settled in their favor in 2005, and of course Apple has its own product.

Differentiation: Rhapsody had the right idea in terms of creating an attractive platform with high quality editorial information, interactive tools and quality content. Acquiring exclusive content is important, as is providing a superior customer experience through user interface and the available of additional content like reviews.

Business Model/Pricing: Driving adoption of your particular business model(s) is essential. For subscription based services, it seems that a sensible strategy would be to give low-cost subscriptions (but not necessarily free, since that tends to make people feel as though it is an inferior product). Getting people to commence a subscription and leave their credit card information helps lock-in the customer as well.

Barriers to Entry: While it is a digital good, a piece of content that would be desirable to users must be produced by an artist, corporation, professional sports league, etc., go through post-product and be marketed. The value chain is not simple and entering the online content market is complex business. It is not easy to set up all the relationships, satisfy all the players and do so on a scale that would be worthwhile

Suggestions for Digital Music Companies

Obviously, this seems a lot easier in retrospect. Some strategic ideas would be:
1) Partner up with hardware makers and other distribution partners to lock-in new users. Do this quickly and beware of being locked out of platforms. Partners like OEMs (Dell, HP), Microsoft, AOL, web portals, etc. would be incredibly important towards acquiring the scale necessary to be a winner.

2) Continue to differentiate by producing quality goods and services around content delivery, and offering higher level packages at a premium. Consider facilitating open innovation and user contributed content to drive with establish base of music enthusiasts to drive differentiation at a low cost. This also helps to amplify the advantages of existing network externalities that RealNetworks has.

3) Get users to try the subscription service at a discount knowing that they can be profitable long term customers with at some switching costs. Meanwhile, make it very easy for users to switch to Rhapsody through ease of use and integration with software and hardware that users are already using.

Saturday, September 13, 2008

Open innovation for health care in the developing world

What do you get when you squish a project with General Electric to transmit ultrasound images in Belize and a project with Center for Infectious Disease in Zambia (CIDRZ) to diagnose cervical cancer in rural areas? I’m finding out over the next three months by using my technical project management skills on an important project to figure out how to transmit images over low bandwidth mobile networks.

The goal is to build a scalable technology to transmit, annotate and manage these records. The key bottleneck in health care in the developing world is human resources. You need an approach for telehealth where image capture can be done by a layman in a rural area. The image can be transmitted and then worked on in a different location (anywhere in the world) where medical expertise is more readily available.

This is necessary for successful preventive screening models in rural areas where clinics might be far from the patients who cannot spend much time away from home. In cervical cancer, one of the biggest problem is loss of patient to follow-up. Near real-time diagnosis using a vinegar swab prevents that as the patient never leaves the clinic while waiting for their diagnosis.

Figuring out some of the key requirements will definitely be a challenge. We’ve heard 100KB or 3MB. That’s a big difference in terms of what type of technology, transmission times, compression, etc. that you would use. We discussed how the key bottleneck in health care in the developing world is human resources. You need an approach for telehealth where image capture can be done by a layman in a rural area. In this case being piloted in Belize exams will be carried out by non-expert sonographers who can send by mobile link to radiologist. DICOM communications server will be used with a database for archiving images and a image reviewer. Currently, GE data system aggregates a series of data, but the process is a bit complicated for a rural health worker. We’ll need to look at this process closely in our needs assessment. The idea is to come up with something that can be scaled and re-used across many developing nations.

Currently, GE has a pilot project going on in Belize. 270 scans were taken in Belize by rural health workers and transmitted using a prototype device. They are conducting full statistical study with panel of five radiologists will score images on whether the quality is sufficient to make diagnoses.

GE’s ultrasound system will address several areas, including pregnant mothers, but also functions like sweeping for typical issues of the thyroid, kidney and gall bladder.

I'm looking forward to learning more about the Zambia process as well - this project is underway as a prototype was developed earlier this year, but the architecture we develop will be important in making it a sustainable model. Here's an overview of the project.

I’m going to be posting out updates throughout the next three months that we’ll be working on this project. Your feedback would be great. This really is the best type of open innovation project. We have leading faculty from MIT, Harvard, one of the top companies in the world in GE and an important non-profit in CIDRZ. The core project team is two bioengineers, an industrial designer from the Media Lab, a programmer and me, the MBA. Let’s see how it goes and if we can help make a big leap scalable across the developing world in improving rural health care.

Bill Lumbergh is a MIT alum?

Just caught this today when I was watching Office Space for the 9,000th time. In one of the scenes with the two consultant Bobs, you catch a glimpse of his bio sheet. I thought I saw MIT on it. When I Googled around, you find out that Lumbergh has a BS and an MS in physics from MIT. Actually, that's pretty interesting. I don't know too many folks who are getting a master's - most people get PhDs. So maybe our friend decided to leave early for the corporate world, or he got the old terminal master's - where they unceremoniously boot you out of the PhD program right after you get your master's.

Wednesday, September 3, 2008

Nynex - ProCD - Digital Directory Assistance Bertrand Pricing War

Just scribbling down this story since I thought it was a good one. Professor Erik Brynjolfsson cited it as an example of Bertrand pricing in our Economics of Information class today.

In 1986, Nynex was creating directories on disk and selling them to companies for approximately 10,000 apiece. A Nynex employee thought this was a pretty good business, so he hired a consultant to start his own business doing it (ProCD). Apparently, the consultant thought it was was a good idea too, so he created Digital Directory Assistance to do the same thing.

Pretty soon, a bidding war broke out with the firms competing on price. This, as Bertrand posits, drove the price down to roughly marginal cost, meaning everyone was pretty much out of luck. Now of course, this info is free on the web.

It's a good illustration of how generally digital information products have high fixed costs and low marginal costs, driving many of the prices towards zero unless there's some type of barrier to entry such as intellectual property.