1
00:00:00,980 --> 00:00:05,840
But today we'll be talking about blockchain and how blockchain can be usefully applied in the context of consent.

2
00:00:07,700 --> 00:00:12,500
Consent. It can be useful both in clinical trials when consent needs to be recorded,

3
00:00:13,100 --> 00:00:20,899
but also and that's what I'll be talking about primarily for managing consent mechanisms for

4
00:00:20,900 --> 00:00:26,990
pre-existing for epidemiology or data driven research that relies on on data that already exists.

5
00:00:29,180 --> 00:00:35,209
So by the turn of the 20th century, the introduction of scientific techniques into medical research was increasing

6
00:00:35,210 --> 00:00:40,190
demands for subjects to study with new knowledge came new hypotheses to be tested,

7
00:00:40,700 --> 00:00:47,330
but subjects were often hard to come by. Therefore, it was tempting for medical researchers to use the resources they had at hand,

8
00:00:47,750 --> 00:00:51,080
namely patients in the hospital that were admitted for other reasons.

9
00:00:52,220 --> 00:00:56,720
From time to time, these patients would suffer some kind of injury from these experiments,

10
00:00:56,720 --> 00:01:02,750
and quite often they would have no clue that these additional procedures were being performed on them,

11
00:01:03,410 --> 00:01:06,530
not for their own benefit, as it were, but as part of an experiment.

12
00:01:07,460 --> 00:01:14,990
And as a result of that public debate concerning the ethics of such research intermittently flared up in the media and in political fora.

13
00:01:16,680 --> 00:01:23,970
In 1898, for example, this man, Albert Nasser, published a series of trials concerning serum therapy in syphilitic patients.

14
00:01:25,270 --> 00:01:28,090
Nasser had attempted to produce a vaccine against the disease.

15
00:01:29,080 --> 00:01:36,580
But to test it out, he went searching for syphilis, free prostitutes at his university hospital, D.C., injected with cell free serum.

16
00:01:37,570 --> 00:01:42,130
And when some of them later went on to contract syphilis, he concluded that his vaccination was ineffective.

17
00:01:43,590 --> 00:01:48,900
Moreover, he argued that they had contracted syphilis because they were prostitutes and not because he had injected them with it.

18
00:01:50,670 --> 00:01:54,809
To local and national media. This is presser and we quickly picked up the story.

19
00:01:54,810 --> 00:02:00,330
And later in the same year, Professor Nisar was fined by the Royal Disciplinary Court for his failure to

20
00:02:00,330 --> 00:02:04,170
obtain consent from his subjects and not notably for the harm he had done.

21
00:02:05,490 --> 00:02:13,530
The next year, 1899, the case prompted the relevant press and minister to commission a report from the Scientific Medical Office of Health,

22
00:02:13,530 --> 00:02:25,550
which included a number of August German researchers as foreman and Vienna, who detailed the story and how they recommended the BMJ article noted.

23
00:02:25,800 --> 00:02:33,240
The Commission focussed explicitly on two important factors when it came to medical the ethical assessment of medical experimentation on humans.

24
00:02:34,260 --> 00:02:37,500
The first was beneficence and the second was autonomy.

25
00:02:38,700 --> 00:02:46,650
They concluded that both informing the patient and getting their consent were both prerequisites for ethical medical experimentation.

26
00:02:48,790 --> 00:02:55,689
These rules were later formalised first in a 1931 pressing government directive, and then, more famously at noon back.

27
00:02:55,690 --> 00:03:01,180
And here we can see Blanche, who was the personal physician to Hitler being given his death sentence.

28
00:03:02,710 --> 00:03:09,550
The prosecution focussed on the coercion rather than the harm that was involved as a primary criterion

29
00:03:09,550 --> 00:03:15,459
for the unethical status of the Nazi doctors experiments during the war over the subsequent decades,

30
00:03:15,460 --> 00:03:22,000
and notably in the sixties and seventies. Laws and policies introduced informed consent requirements for most human subjects.

31
00:03:22,000 --> 00:03:29,950
Research in most places that such research was carried out. By the turn of the 21st century,

32
00:03:30,250 --> 00:03:33,969
the introduction of information and computing technologies into medical research

33
00:03:33,970 --> 00:03:37,510
was causing an increase in both the supply and demand of biomedical research data.

34
00:03:38,530 --> 00:03:44,649
But concerns about consent were by now taking a different turn. For example, in 2003,

35
00:03:44,650 --> 00:03:50,559
the results of a survey conducted by the Association of American Medical Colleges found that consent regulations in

36
00:03:50,560 --> 00:03:57,640
the form of the US Privacy Rule had won reduced patient recruitment to increase the likelihood of selection bias.

37
00:03:58,300 --> 00:04:02,230
Three Increase the cost of conducting research by requiring more paperwork, yadda yadda.

38
00:04:03,010 --> 00:04:07,960
Four Increase the number of errors in research. Or de-identified information was being used.

39
00:04:08,680 --> 00:04:13,810
And five made multi-site trials more difficult because of variations in local IAB review.

40
00:04:14,820 --> 00:04:22,870
And finally, Six had caused researchers to abandon projects entirely because of an increased number of rules for operating a research study.

41
00:04:24,340 --> 00:04:28,690
The survey, along with many others other sources of evidence,

42
00:04:28,690 --> 00:04:34,570
is quoted approvingly by the U.S. National Academies Institute of Medicine Report 29, which you can see here.

43
00:04:36,220 --> 00:04:42,760
By this time, it was becoming increasingly clear that rules made to protect prostitutes from random syringes and any

44
00:04:42,760 --> 00:04:48,880
human against medical war crimes might not be necessarily appropriate for all other kinds of research.

45
00:04:49,900 --> 00:04:56,200
In particular, these rules are not tailored to nor flexible towards research based on pre-existing data.

46
00:04:57,130 --> 00:05:00,970
For example, in electronic health records and other medical databases.

47
00:05:02,190 --> 00:05:03,900
With the exception of the Nordic countries,

48
00:05:04,410 --> 00:05:11,370
which have allowed registry based research for health data and many other kinds of data since the sixties, largely for economic competitive reasons,

49
00:05:12,960 --> 00:05:20,160
countries around the world have largely failed to update their consent legislation to accommodate the changes over the last 50 years,

50
00:05:20,580 --> 00:05:24,120
particularly failed to accommodate data driven medical research.

51
00:05:26,950 --> 00:05:31,420
This is a problem because these rules were made to protect human subjects in the flesh,

52
00:05:32,350 --> 00:05:37,300
and they did not foresee the very different issues involved when it comes to the use of people's data.

53
00:05:38,740 --> 00:05:46,810
Although asking people for consent to use their medical data respects their autonomy in much the same way as asking for consent before interventions.

54
00:05:47,860 --> 00:05:51,100
The potential for harm. And thus the stakes are radically different.

55
00:05:52,270 --> 00:05:56,280
Smaller. At the same time, the potential for beneficence,

56
00:05:56,700 --> 00:06:03,540
the for the use of these medical data to lead to a generally better world for everyone has increased appreciably.

57
00:06:05,520 --> 00:06:12,630
So why would anyone argue against consent requirements beyond being enshrined in various codes and laws?

58
00:06:12,840 --> 00:06:14,940
Consent requirements make intuitive sense.

59
00:06:16,170 --> 00:06:22,020
If you're going to expose someone to potential harm for the benefit of others, then you better make sure that they're game for it.

60
00:06:23,970 --> 00:06:33,390
But there are good reasons. One major reason is the difficulty in obtaining consent from individuals in a registry or database setting database study.

61
00:06:34,500 --> 00:06:42,390
For example, many individuals in our registry might be deceased or they might have changed address making them difficult to contact,

62
00:06:43,200 --> 00:06:49,710
or they may miss or ignore or simply not care about letters, phone calls, emails, and other requests for consent.

63
00:06:51,170 --> 00:06:53,780
In one extreme case mentioned in the Institute of Medicine report,

64
00:06:53,780 --> 00:07:01,490
it quoted earlier registration to the Hamburg cancer registry fell by 70% after consent requirements were introduced,

65
00:07:01,850 --> 00:07:10,430
after which use of the registry had to be discontinued. Although this percentage is going to differ widely in various cases,

66
00:07:10,940 --> 00:07:16,910
it is clear that consent requirements sometimes lead to reduced subject recruitment, just as the Institute of Medicine said.

67
00:07:18,340 --> 00:07:22,480
This in turn, leads to economic costs and delay in the conduct of research,

68
00:07:22,870 --> 00:07:28,000
which translates into ailments not ameliorated and at the extremity labs not safe.

69
00:07:29,710 --> 00:07:35,290
More significant still, is the impact that these consent requirements have on the quality of medical research that is conducted.

70
00:07:36,240 --> 00:07:41,670
We cannot generalise the results of the study if the underlying population is not representative.

71
00:07:42,210 --> 00:07:46,050
For example, if it is carried out only on young or elderly people.

72
00:07:47,230 --> 00:07:52,480
Similarly, research using volunteer data will yield results that are not valid for non

73
00:07:52,480 --> 00:07:57,700
volunteers because volunteers and non-violence use differ in important respects.

74
00:07:59,620 --> 00:08:06,640
One meta analysis on this topic, which is called Selection Bias, looked at nine variables known to affect responses to treatment.

75
00:08:07,960 --> 00:08:12,760
All nine of these variables differed significantly between considers and none considers.

76
00:08:13,690 --> 00:08:17,020
The extent of the bias varied in each case to.

77
00:08:18,180 --> 00:08:21,870
Worse, none of these biases could be mathematically corrected.

78
00:08:22,910 --> 00:08:30,770
Thus wherever consent is required, this is a statistically irremediable bias of unknown magnitude and significance automatically appears.

79
00:08:32,650 --> 00:08:37,450
There are also more fundamental problems for consent requirements as judicially conceived.

80
00:08:38,620 --> 00:08:39,880
In order to be legitimate,

81
00:08:40,900 --> 00:08:47,680
the person consenting must understand and assess the most relevant risks and benefits likely to be involved in the specific research study.

82
00:08:48,820 --> 00:08:55,570
This is difficult to do in 1899. Nowadays, few comprehend the details behind a biomedical study outside their speciality.

83
00:08:56,730 --> 00:09:02,130
The researchers conducting the study may themselves not even know all the relevant risks or benefits involved.

84
00:09:03,750 --> 00:09:07,230
Consent is also supposed to be specific to a singular instance.

85
00:09:08,040 --> 00:09:16,110
Yet with changes in computing and information technology, the same person's data may be queried hundreds and hundreds or thousands of times each year.

86
00:09:17,550 --> 00:09:21,720
We really study the minutiae of the protocol each time it is tweaked.

87
00:09:22,790 --> 00:09:26,270
And do we even have the time and motivation to do so if we want it?

88
00:09:28,320 --> 00:09:35,770
For example. McDonald and Crane were found in 2008 that simply reading never mind understanding

89
00:09:36,370 --> 00:09:40,330
the privacy policies that an average American was likely exposed to online.

90
00:09:40,900 --> 00:09:50,890
In a year it would have taken 76 working days. This is simply not a scalable option as the complexity and volume of research continues to increase.

91
00:09:53,130 --> 00:10:00,060
In 2016, Julian said. The rescue by Sahakian and I argued that a duty of easy rescue applies to medical

92
00:10:00,060 --> 00:10:04,680
data contribution because contributing data does not involve significant risks,

93
00:10:04,830 --> 00:10:12,090
costs or effort. It has large potentially life saving benefits and refraining from from sharing

94
00:10:12,600 --> 00:10:15,870
seriously hinders the provision of these benefits to individuals and groups.

95
00:10:15,870 --> 00:10:22,980
There are no other ways to get there. This concept of minimal risk is important for the discussion I want to have today,

96
00:10:23,460 --> 00:10:26,820
because we believe that it might be possible to reduce the risk of most,

97
00:10:27,270 --> 00:10:33,300
if not all, records based research to a minimum using advances in computing and encryption.

98
00:10:34,120 --> 00:10:38,560
Notably a set of technologies known as blockchain and secure multiparty computing.

99
00:10:40,210 --> 00:10:46,460
So what is a blockchain? A blockchain is essentially a better way to record transactions.

100
00:10:46,850 --> 00:10:51,680
We might say that a transaction occurs when X gives A receives Y from Z.

101
00:10:52,700 --> 00:10:58,520
Here X and Z could be one or more entities of any kind, say one or more humans or computers.

102
00:10:59,580 --> 00:11:05,610
And why is the thing that is transacted could be data or items of any kind and description and quantity.

103
00:11:06,850 --> 00:11:10,900
To illustrate the first application of blockchain technologies with the Bitcoin currency.

104
00:11:12,660 --> 00:11:16,500
And this concerns monetary transactions, mostly between individuals.

105
00:11:17,550 --> 00:11:22,680
So in this case, X and Z are individuals or groups of humans or perhaps even institutions.

106
00:11:24,150 --> 00:11:31,770
And they are involved in a mutual transaction transaction of WHI, which is some quantity of fiscal value here represented in the Bitcoin currency.

107
00:11:33,620 --> 00:11:36,890
As you might imagine by analogy to fiat money, to normal money,

108
00:11:37,640 --> 00:11:43,160
many such transactions take place during any given window of time to make things more manageable.

109
00:11:43,190 --> 00:11:52,670
These transactions can be grouped together periodically and aggregated into chunks or blocks of transaction data indexed by time.

110
00:11:55,050 --> 00:12:00,330
The Bitcoin transaction data in block one precedes the transaction data in block two

111
00:12:01,470 --> 00:12:05,970
and the two blocks are chained together by their sequential release and in time.

112
00:12:07,570 --> 00:12:11,780
Each block carries a timestamp and a hash of the previous block.

113
00:12:11,800 --> 00:12:20,990
Now, a hash is essentially a mathematical function where you give it a certain input and it will come out with a certain output.

114
00:12:21,010 --> 00:12:24,220
But if you change the input in any way, the output would be radically different.

115
00:12:24,850 --> 00:12:31,270
So for example, if you have downloaded illegal materials like a movie and you want to check whether there is a virus in it,

116
00:12:31,930 --> 00:12:37,920
you could go online and find out what data stream of actual movie should has to hash

117
00:12:37,930 --> 00:12:41,700
the file that you have downloaded and see whether it comes out with the same value.

118
00:12:41,710 --> 00:12:45,340
If not, then it's been messed with. It's also used in many other things.

119
00:12:50,590 --> 00:12:55,060
The database, which stores this information is not kept at headquarters,

120
00:12:55,660 --> 00:13:01,309
but crucially it is distributed amongst hundreds of thousands of nodes and, uh, server nodes.

121
00:13:01,310 --> 00:13:05,080
So node is a computer or any computing device.

122
00:13:06,280 --> 00:13:12,040
In fact, anyone who has a reasonably young computer or a reasonably new smartphone can download

123
00:13:12,040 --> 00:13:15,910
and store this database and run it in its entirety since its since inception.

124
00:13:16,990 --> 00:13:19,150
This has a number of important consequences.

125
00:13:20,720 --> 00:13:28,130
For one, the structure makes the blockchain resilient against attacks because the record is distributed amongst hundreds of thousands of nodes.

126
00:13:28,670 --> 00:13:33,560
You get one. You need to get the others for a new block to be added to the chain.

127
00:13:34,370 --> 00:13:39,440
A majority of these servers are nodes. Have to agree on the contents.

128
00:13:40,940 --> 00:13:47,900
Thus the computer which proposed a block in which there was a transaction giving user x say access to the funds or information of prison

129
00:13:48,080 --> 00:13:57,230
officers and said this will achieve nothing unless at least 50% of the other computers are persuaded or verify that this is indeed the case.

130
00:13:58,470 --> 00:14:06,210
Similarly, if someone attempts to change the previous block, this will cause that block and all subsequent blocks to change their hash values.

131
00:14:06,750 --> 00:14:11,130
And this is a clear signal that someone has tried to tamper with your record.

132
00:14:13,080 --> 00:14:18,000
The system is also set up in such a way that hacking or controlling that many computers

133
00:14:18,360 --> 00:14:23,700
would cost more in electricity consumption than the resulting outcome would be worse,

134
00:14:24,630 --> 00:14:30,900
even if it were possible to do a hostile takeover thanks to something known as the proof of work protocol.

135
00:14:32,230 --> 00:14:38,590
Um. This is what is meant when people say that blockchains are tamper proof.

136
00:14:40,580 --> 00:14:46,010
Moreover, each entity in a blockchain transaction is represented by a cryptographic identifier.

137
00:14:46,990 --> 00:14:55,060
And it is can choose whether to exchange that identifier for each transaction, thus maximising their level of pseudo anonymity.

138
00:14:55,930 --> 00:15:02,710
Or they could choose to use the same identifier repeatedly, thus opting for less pseudo anonymity.

139
00:15:06,610 --> 00:15:15,790
To access the money data. Information that is stored or associated with an identifier requires a corresponding password known as a private key.

140
00:15:16,030 --> 00:15:20,380
In addition to the public key, which from which the identifier can be generated.

141
00:15:21,860 --> 00:15:23,120
The details here are not important.

142
00:15:24,290 --> 00:15:31,310
Using a private key, an individual can create a unique digital signature for any message or transaction, thus verifying their identity.

143
00:15:31,760 --> 00:15:40,139
No one else could do so. Anyone who possesses a corresponding public key can use that key to verify that that

144
00:15:40,140 --> 00:15:45,720
message did indeed come from the person represented by the private key signature.

145
00:15:48,200 --> 00:15:50,450
Both of these keys are created through an iterated one way,

146
00:15:50,450 --> 00:15:55,400
hashing out a rhythm which makes it impossibly difficult to guess or reverse engineer the private key.

147
00:15:56,290 --> 00:16:00,160
Knowing only the public key or any other information except for the private key.

148
00:16:01,300 --> 00:16:07,270
Thus, even if the blockchain were to be compromised by a malicious actor with more than 50% stake of service under their belt,

149
00:16:07,990 --> 00:16:11,590
they would not be able to figure out who all these people are.

150
00:16:13,060 --> 00:16:16,720
Indeed, the hope Bitcoin blockchain is transparent because of this feature.

151
00:16:17,230 --> 00:16:20,860
Anyone can see the entirety of history of transactions who sent how much to whom?

152
00:16:20,860 --> 00:16:25,090
When? Because no one knows who that is. And the stats are.

153
00:16:27,080 --> 00:16:32,230
Unless they want to be known. So that is the blockchain in a nutshell.

154
00:16:32,590 --> 00:16:38,170
It's, of course, a lot more complicated than that. And there are lots of variations and numerous modifications.

155
00:16:39,400 --> 00:16:47,240
For example, it is possible to add code to a blockchain or layer on top of the blockchain, which can trigger further events to happen.

156
00:16:47,260 --> 00:16:50,080
If certain criteria, predefined criteria are met.

157
00:16:51,580 --> 00:16:58,720
Such pieces of code are known as smart contracts and they can be used to automatically fulfil transactions or carry out calculations.

158
00:16:59,050 --> 00:17:04,060
Once these predefined criteria have been met. So a classic example. I want Doug to give me a house.

159
00:17:04,630 --> 00:17:07,390
Dog wants to get money for the house, but he's not sure I won't run off.

160
00:17:07,990 --> 00:17:15,220
So we create a smart contract that says once the criterion of house built is fulfilled, the money will automatically leave my account and go to dog.

161
00:17:21,460 --> 00:17:29,260
Crucial thing, though, about all this is that the blockchain can record any type of data, not just Bitcoin, finance, whatever.

162
00:17:30,430 --> 00:17:37,480
It could just as easily store access keys to electronic health records or record the sequential steps in a clinical trial protocol.

163
00:17:38,430 --> 00:17:46,950
Or even buy a smart contract, automate child data release and integrate such child data into near real time meta analysis.

164
00:17:48,010 --> 00:17:54,669
And it can do this safely, quickly, transparently, efficiently, with effectively zero chance of privacy breach,

165
00:17:54,670 --> 00:17:58,840
effectively zero cost, and no chance at all of transaction falsification.

166
00:18:01,320 --> 00:18:10,530
For example, the Enigma Project uses blockchain technologies to manage access to data which are themselves stored in a location not on the blockchain.

167
00:18:11,880 --> 00:18:13,800
Data are encrypted upon collection,

168
00:18:13,800 --> 00:18:22,470
using an encryption key that is shared between the data owners and the data requires only a hash of this original data is kept on chain.

169
00:18:23,840 --> 00:18:29,239
Computations can then be performed on small chunks of still encrypted data distributed throughout the Enigma

170
00:18:29,240 --> 00:18:35,750
Network so that no one at any point can tell either the pseudo identity of the data owners or acquirers,

171
00:18:36,410 --> 00:18:43,830
nor the data itself. This data can then be queried by subject and investigator in aggregated form,

172
00:18:43,830 --> 00:18:49,640
with statistics performed whose identities are in turn verified by the encryption keys.

173
00:18:51,720 --> 00:18:53,600
In another example, Nebula Genomics.

174
00:18:54,390 --> 00:19:02,310
A person wishing to access data since a request by a blockchain transaction to relevant nodes in the network once authenticated,

175
00:19:02,310 --> 00:19:09,000
the subset of data relevant to the researchers study is then sent automatically via a smart contract to a data enclave,

176
00:19:09,570 --> 00:19:17,160
at which point it is de-identified to the highest level of abstraction compatible with research aims and aggregated before being made available.

177
00:19:18,780 --> 00:19:23,220
Of course, there are many other models and potential ways to do this.

178
00:19:24,950 --> 00:19:32,600
In addition, the blockchain enables novel interactions between all kinds of stakeholders in the health care environment.

179
00:19:33,870 --> 00:19:41,520
Whereas your research. Whereas previously a doctor would ask an individual for consent to take observations or to share data.

180
00:19:42,300 --> 00:19:48,090
It is possible through the blockchain for that or any other individual to request data or

181
00:19:48,090 --> 00:19:52,649
otherwise interact with the researcher the other way around at any level of anonymity.

182
00:19:52,650 --> 00:20:01,170
Pseudo anonymity. This feature, which we termed Frozen, opens up a veritable cornucopia for health data sharing.

183
00:20:02,630 --> 00:20:08,990
On one level, it could be envisaged as a data marketplace in which patients with rare genomes can sell

184
00:20:08,990 --> 00:20:15,650
their genetic data to pharmaceutical companies or university researchers at high prices.

185
00:20:17,060 --> 00:20:23,630
At another it kipping considered a marketplace of ideas in which patients with rare disorders could seek out one another.

186
00:20:24,580 --> 00:20:30,400
We'll offer researchers incentives to study their conditions, which are otherwise not being offered.

187
00:20:32,050 --> 00:20:42,190
Similarly, the government or any government or really any institutional or individual or computer could in theory become involved

188
00:20:42,340 --> 00:20:49,510
with any research project through this present mechanism without anybody necessarily knowing who the other person is,

189
00:20:49,510 --> 00:20:56,320
except that they are who they claim to be. So for example, I claim to be a medical researcher.

190
00:20:56,350 --> 00:21:00,190
You can verify that, but you will not find out. My name is so.

191
00:21:02,800 --> 00:21:04,990
But here again, there are critical questions.

192
00:21:05,980 --> 00:21:12,700
What kinds of data should be up for sale and which should only be allowed to be donated and which shouldn't even be allowed to be done?

193
00:21:14,360 --> 00:21:19,010
Should there be restrictions on what kind of data people can buy without a background in health research?

194
00:21:20,550 --> 00:21:27,180
What criteria would qualify one as a researcher allowed to access some types of sensitive data through this mechanism?

195
00:21:28,630 --> 00:21:36,940
Which criteria would disqualify an individual or organisation from donating or acquiring or buying or selling data?

196
00:21:40,030 --> 00:21:44,650
So blockchain is definitely something I think that we should be thinking about when it comes to the use of health data.

197
00:21:45,520 --> 00:21:49,210
The interesting question is not whether but how we should implement it.

198
00:21:51,070 --> 00:21:55,360
One attractive proposal is to give each individual access to their own health data.

199
00:21:56,650 --> 00:22:04,300
The health data would be stored somewhere secure and in the cloud of a physical location, and the key to access it would be stored on a blockchain.

200
00:22:05,220 --> 00:22:13,170
This key could then be given by its possessor to their physician or really any of anybody else that they would allow access to,

201
00:22:13,380 --> 00:22:20,090
that they would like to have access to their records. All kinds of access management would be possible here in theory.

202
00:22:20,620 --> 00:22:27,430
There could be keys to the whole record. Accused to part of the record. Single use, multi-use keys, you name it.

203
00:22:29,440 --> 00:22:33,790
Using an API and application programming interface designed for ease of use.

204
00:22:34,180 --> 00:22:41,560
Data owners could grant, modify or revoke permissions to access data by means of a near instantaneous transaction on the blockchain.

205
00:22:43,250 --> 00:22:50,620
In doing so, they still retain access to their own data. But gain greater flexibility in determining who else may use it.

206
00:22:52,180 --> 00:22:57,130
For example, some data owners may wish to grant permission for data access only to qualified researchers.

207
00:22:58,200 --> 00:23:02,220
We're only two researchers working on publicly funded projects or from public institutions.

208
00:23:04,610 --> 00:23:09,530
They may also like in the Nebula model briefly discussed earlier make their data available to

209
00:23:09,530 --> 00:23:14,540
those willing to purchase it or to a combination of these modalities or really any other.

210
00:23:15,770 --> 00:23:22,490
Alternatively, they may make their data completely open for anyone to see, like their research and his Open Humans Project.

211
00:23:22,790 --> 00:23:29,730
Personal Genome Project. Or they can keep exes entirely restricted to themselves, not even letting their doctor see.

212
00:23:31,730 --> 00:23:36,740
Importantly, this would allow people to treat different kinds of data differently.

213
00:23:36,950 --> 00:23:43,579
Medical data, for example, access to sensitive medical data, which may concern issues such as reproductive choices,

214
00:23:43,580 --> 00:23:52,400
mental health history, substance use services and so on, might be kept entirely private, but granted only to highly select entities.

215
00:23:53,810 --> 00:24:02,210
Other types of medical data, such as flu status or height, could be put up for sale for donation to anyone that wishes to use.

216
00:24:03,760 --> 00:24:12,400
And this way the blockchain would provide a practical means of implementing the notion of matter concerns introduced by some sometime in 2016.

217
00:24:13,890 --> 00:24:17,280
The individual would have maximum control in this case over their health data.

218
00:24:18,000 --> 00:24:21,810
The ethical principle of autonomy would correspondingly be maximised.

219
00:24:23,580 --> 00:24:28,200
Alternatively, however, one could imagine other ways of arranging access control.

220
00:24:29,070 --> 00:24:33,540
For example, some information on a person's health record might simply not be that sensitive.

221
00:24:34,860 --> 00:24:40,919
That type of information could be grouped together, and then access key to non-sensitive information might be made automatically available

222
00:24:40,920 --> 00:24:46,140
to anybody who fulfils certain criteria and qualifications as a researcher.

223
00:24:47,370 --> 00:24:52,950
Or I might go one step further and give qualified researchers access privileges to all health data relevant to their study.

224
00:24:53,790 --> 00:25:01,160
Of course, the form of this data would reach. The researchers then would be aggregated de-identified with all the statistics already done correctly.

225
00:25:02,280 --> 00:25:09,030
So did the smart contract. Thus, no individual identifiable information would be made available to any human eye.

226
00:25:10,650 --> 00:25:17,820
Or really to any computer either because it's encrypted. When they perform a calculation that the difference such a system could make

227
00:25:18,510 --> 00:25:22,320
to the efficacy and efficiency of biomedical research is hard to overstate.

228
00:25:24,940 --> 00:25:31,420
So there are these options. And the question is, how should we choose between them?

229
00:25:32,140 --> 00:25:36,280
Well, the question is which one should we choose, but also what kinds of reasons or thoughts?

230
00:25:37,060 --> 00:25:43,190
How should we be deciding this? We should be deciding. One one way.

231
00:25:44,680 --> 00:25:54,850
Is my favourite bioethical analysis. So the easy rescue that I introduced earlier, strongly motivated towards some kind of mandated data sharing.

232
00:25:56,030 --> 00:25:59,450
If the blockchain could make privacy breaches and hacks pretty much impossible,

233
00:25:59,960 --> 00:26:10,130
then it seems to me that there are no or only minimal risks involved in my data being used for medical research on encrypted data.

234
00:26:10,340 --> 00:26:19,200
Encrypted data. And if I did it. And because of this, it seems to me that we should that I said that we should, everybody should.

235
00:26:20,320 --> 00:26:26,400
And furthermore, that it would not be wrong to require people to do so passively if there are no burdens to them.

236
00:26:27,480 --> 00:26:31,290
So that's my personal view. I lean towards beneficence.

237
00:26:31,890 --> 00:26:38,610
And in this in this case, just regarding people's autonomy so as to increase justice and for the greater good,

238
00:26:38,610 --> 00:26:42,300
seems like a good weighing up of the principles.

239
00:26:43,320 --> 00:26:46,650
But of course, reasonable people disagree, and I would love to talk to anybody that does.

240
00:26:48,600 --> 00:26:51,540
Another way to approach these issues is through public engagement.

241
00:26:52,740 --> 00:26:57,600
The Danish Board of Technology, for example, has done excellent work on contentious scientific issues,

242
00:26:57,600 --> 00:27:02,370
such as genetically modified genetic modification of various types of stem cell research.

243
00:27:03,480 --> 00:27:04,590
In one of their models.

244
00:27:04,590 --> 00:27:15,480
They ask citizen volunteers to meet every weekend for some months, and then they are given scientific lectures and homework by both pro and con.

245
00:27:16,380 --> 00:27:20,010
And in the final by the final weekend, they have a deadline and they produce a report.

246
00:27:20,010 --> 00:27:23,580
And this report is then considered to be the considered judgement of the Danish people.

247
00:27:27,110 --> 00:27:30,200
Of course, there are many other ways to do public engagements,

248
00:27:30,380 --> 00:27:36,500
but some public engagements should be absolutely or would be absolutely necessary

249
00:27:36,500 --> 00:27:39,860
to achieve legitimacy for any change in the way that we handle people's data.

250
00:27:39,890 --> 00:27:42,590
Just think about what happens with health data.

251
00:27:45,410 --> 00:27:52,910
Although public engagement is fascinating, my colleagues and I are currently interested in a very different third way,

252
00:27:53,900 --> 00:27:58,490
namely the human right to enjoy the benefits of the progress of science and its applications.

253
00:28:00,020 --> 00:28:06,889
This human rights is much neglected, but it has the same legal status as all the other economic,

254
00:28:06,890 --> 00:28:10,670
social and cultural rights enshrined in the Autonomous International Covenant.

255
00:28:12,440 --> 00:28:17,990
This means that just like governments are required to respect, say, the right to health or free assembly,

256
00:28:18,710 --> 00:28:23,870
so they must also respect the right to enjoy the benefits of scientific progress and its applications.

257
00:28:25,730 --> 00:28:30,320
The right to science is highly underdeveloped, but international human rights law,

258
00:28:30,320 --> 00:28:35,480
which is used to interpret other human rights obligations, is equally applicable to it.

259
00:28:36,590 --> 00:28:44,450
Thus, we can argue by analogy, for example, to the right to health. The governments must ensure that as many individuals as possible benefit as

260
00:28:44,450 --> 00:28:48,799
much as possible from the advances that are made in science and technology,

261
00:28:48,800 --> 00:28:52,490
as long as any policies to that effect are also human rights compliance.

262
00:28:54,850 --> 00:28:56,950
One such argument could be quite simple to make.

263
00:29:00,240 --> 00:29:04,620
Since improved medical research constitutes a benefit of the progress of science and has applications.

264
00:29:04,890 --> 00:29:11,010
The starting point from a human rights perspective is that everyone should benefit, where possible, from improved medical research.

265
00:29:12,430 --> 00:29:15,670
Like other human rights, though, this rights is not absolute.

266
00:29:16,010 --> 00:29:21,400
It may be outweighed by other rights or concerns or the underlying principles in human rights.

267
00:29:22,150 --> 00:29:25,030
However, any such limitations must be appropriate.

268
00:29:25,720 --> 00:29:33,130
They must be proportionate, and they must be strictly necessary for the general well-being in a democratic society.

269
00:29:34,000 --> 00:29:39,700
Otherwise they are unlawful. We discuss this further for those that are interested in the articles shown here.

270
00:29:41,110 --> 00:29:44,620
So the Right to Science Framework thus provides us, I think,

271
00:29:45,340 --> 00:29:50,740
with an approach to the question of consent in medical research that also has some legal weight added to it.

272
00:29:52,060 --> 00:29:55,870
Roughly, we could argue from a right to science perspective that blockchain be used to facilitate

273
00:29:56,110 --> 00:30:00,339
medical research to the extent that this is necessary for the general well-being,

274
00:30:00,340 --> 00:30:08,020
which I believe it is, but only as long as this is consistent with other human rights and is appropriate and is proportionate.

275
00:30:09,030 --> 00:30:15,629
We might then discuss whether some kinds of mandates as automatic data processing are appropriate and appropriate, for example,

276
00:30:15,630 --> 00:30:22,590
where someone does not wish on religious or normative grounds that their data be used for research into abortion or blood transfusions.

277
00:30:23,970 --> 00:30:28,950
We might also wonder whether the full removal of autonomy from consideration is proportionate.

278
00:30:29,700 --> 00:30:33,329
For example, by observing that some low level of selection bias is acceptable.

279
00:30:33,330 --> 00:30:40,970
After all, we have high levels right now. As well as noting that where the feasibility and scientific merit of a proposed data usage is low,

280
00:30:41,660 --> 00:30:45,050
concerns about autonomy become correspondingly weightier.

281
00:30:46,700 --> 00:30:54,200
This is exciting work for me because of all the possibilities that are novel and powerful and for the same reasons.

282
00:30:54,200 --> 00:30:59,659
It is important that we ponder and discuss them, which I propose we do now or shortly after.

283
00:30:59,660 --> 00:31:05,030
I thank the teams involved. This is the blockchain team.

284
00:31:06,900 --> 00:31:11,590
And this human rights. The Science group which originated.

285
00:31:13,480 --> 00:31:17,850
That's Harvard Medical School Centre for Bioethics. And it's still going strong.

286
00:31:19,890 --> 00:31:20,460
So thank you.