contentSlug

type

ContentSlug_ContentQuery

slug

ContentPageContentQuery

content

trendingContent

query ContentSlug_ContentQuery(
  $contentSlug: String!
  $type: [ContentType!]
) {
  ...ContentPageContentQuery_13QmVO
}

fragment CompanyDetails_Content on Content {
  __isContent: __typename
  companies {
    name
    description
    pageUrl
    id
  }
}

fragment ContentBasicInfo on Content {
  __isContent: __typename
  id
  type
  title
  lang
  companies {
    viewerIsFollowing
    id
    name
    pageUrl
  }
  tags {
    viewerIsFollowing
    id
    name
    pageUrl
  }
  ...PublishingDetails
  ...Reactions
}

fragment ContentMetadata on Content {
  __isContent: __typename
  type
  title
  lang
  pageUrl
  alternatePageUrls: translations {
    __typename
    pageUrl
    lang
  }
  imgUrl
  description
  ... on Video {
    liveTime
    videoUrl
  }
  ... on Webcast {
    liveTime
    videoUrl
  }
}

fragment ContentPageContentQuery_13QmVO on Query {
  content(slug: $contentSlug, type: $type) {
    __typename
    id
    type
    title
    isPremium
    lang
    htmlBody
    summary
    companies {
      forum {
        latestPosts {
          ...ForumPosts
        }
      }
      name
      id
      viewerIsFollowing
      pageUrl
      defaultInstrument {
        isin
        id
        marketOpen
        marketClose
        marketplace {
          region
        }
        returns {
          total
        }
      }
    }
    tags {
      viewerIsFollowing
      id
      name
      pageUrl
    }
    createdAt
    ... on Video {
      liveTime
    }
    ... on Webcast {
      liveTime
    }
    slug
    ... on CompanyReport {
      reportUrl
    }
    ... on ThirdPartyCompanyReport {
      reportUrl
    }
    ... on StockExchangeRelease {
      reportUrl
    }
    ...ContentBasicInfo
    ...ContentMetadata
    ...PriceAndRecommendation_Content
    ...CompanyDetails_Content
    ...ReportLink_Content
    ...KeyFigures_Content
  }
  ...TrendingContent_1jQ2Cn
}

fragment ForumPosts on ForumPost {
  text
  likesCount
  url
  author
  postedAt
  authorImgUrl
}

fragment KeyFigures_Content on Content {
  __isContent: __typename
  companies {
    defaultInstrument {
      estimateData {
        modelType
        reportingCurrency
      }
      snapshot {
        transaction {
          date
          actualized {
            year
            quarter
            revenue
            revenueGrowth
            ebitAdjusted
            ebitReported
            ebitda
            enterpriseValue
            epsAdjusted
            dividend
            dividendYield
            sharePrice
          }
          estimates {
            year
            quarter
            revenue
            revenueGrowth
            ebitAdjusted
            ebitReported
            ebitda
            enterpriseValue
            epsAdjusted
            dividend
            dividendYield
            sharePrice
          }
        }
      }
      id
    }
    id
  }
}

fragment PriceAndRecommendation_Content on Content {
  __isContent: __typename
  createdAt
  companies {
    name
    pageUrl
    defaultInstrument {
      priceData {
        tradeCurrency
      }
      snapshot {
        priceData {
          date
          lastPrice
          tradeCurrency
        }
        transaction {
          date
          recommendation
          targetPrice
        }
        prevTransaction {
          recommendation
        }
      }
      id
    }
    id
  }
}

fragment PublishingDetails on Content {
  __isContent: __typename
  createdAt
  publishers {
    name
    imgUrl
  }
  authors {
    __typename
    id
    name
    slug
    imgUrl
    jobTitle
  }
  ... on Video {
    liveTime
  }
  ... on Webcast {
    liveTime
  }
}

fragment Reactions on Content {
  __isContent: __typename
  id
  reactionGroups {
    reactionType
    reactionsCount
    viewerReacted
  }
}

fragment ReportLink_Content on Content {
  __isContent: __typename
  type
  isPremium
  ... on CompanyReport {
    reportUrl
  }
  ... on ThirdPartyCompanyReport {
    reportUrl
  }
  ... on StockExchangeRelease {
    reportUrl
  }
}

fragment TrendingContent_1jQ2Cn on Query {
  trendingContent(type: ARTICLE) {
    __typename
    id
    title
    lang
    pageUrl
    createdAt
    type
    imgUrl
    companies {
      squareLogoUrl
      name
      id
    }
    authors {
      __typename
      imgUrl
      id
    }
    publishers {
      imgUrl
    }
  }
}


__typename

title

isPremium

lang

htmlBody

summary

companies

createdAt

publishers

authors

reactionGroups

pageUrl

translations

imgUrl

description

Head of Nordic ER Development, Analyst

Antti Luiro

Analyst

Frans-Mikael Rostedt

Aiforia Technologies equips pathologists and researchers in preclinical and clinical laboratories with software to translate images into discoveries, decisions and diagnoses. The company's products and services are used for medical image analysis, across a variety of fields such as oncology and neuroscience. Aiforia Technologies is headquartered in Finland.

Does anyone know if Aiforia reports deals obtained through partners? The CEO suggested that a few deals through Siemens would come through before the end of the year.

Yeah, the problem with these health-sector models seems to be that they are always tuned as much as possible, “not so much generalization.” This then causes the model to react very strongly to every small change in the input, even if fine-tuning or changes to the downstream head are made. Fine-tuning should indeed help with this problem (because an old model is being trained with new types of data from a new device), but I haven’t looked into the paper. One solution could also be some kind of encoder at the front end of the data, an “embedding layer,” which transforms the input data into a form that no longer contains scanner-dependent features. All of these are being hotly debated there.

Interesting observation: If a foundation model is created with a specific scanner, the model learns the scanner’s characteristics, but may not necessarily work with other scanners. https://arxiv.org/pdf/2507.22092 The FDA has been right in this regard, that model approval is scanner-specific, even though at the time of the guideline, there probably wasn’t much information about foundation models yet, as they have only been in research use for 1-2 years. The FDA’s interpretations of device requirements seem to be gradually easing. The utilization of foundation models quickly came to Aiforia. Now AI can accelerate model training, and pathologists spend less time on training. The ultimate goal was to use less of a clinical pathologist’s time. Now, that can also be leveraged. And the model can even teach pathologists to see important signals that were previously unknown. This is “only” a vision model. The next step could be multimodal, where images and text, i.e., reports, even medical records, if access were possible, are combined into the model. Perhaps, and someday. Aiforia’s approach is currently the best in terms of control, first task-specific models and always human in the loop. From there, forward with development. If Aiforia’s emphasized robustness, including scanner diagnostics, is included in those vast datasets in foundation models and the patent application, they have overcome the pitfalls of that article. If Aiforia can demonstrate to the FDA that their models have quality, i.e., error tolerance, at a level where all scanners work, then problem solved. Of course, it’s a race, and there are other players in the field.

Here are Antti’s and Frans’s comments on this recent AI platform news.  Aiforia announced on Wednesday that it is launching a new generation AI platform that utilizes Vision Transformer technology. The new technology promises to accelerate the development of AI models and improve their performance, which supports the expansion of the company’s product portfolio. The news does not have a direct impact on our forecasts, but we consider the enhanced product development positive, as it partly mitigates risks regarding the company’s innovation capability after cost cuts and supports the company’s long-term competitiveness.                   Inderes              Aiforia lanseeraa uuden tekoälyalustan - Inderes    Uusi teknologia lupaa nopeuttaa tekoälymallien kehitystä ja parantaa niiden suorituskykyä, mikä tukee yhtiön tuoteportfolion laajenemista.

New technology promises to accelerate the development of AI models and improve their performance, which supports the expansion of the company’s product portfolio. In addition to these good releases, I’d also like to see some contract news.

Aiforia launches next-generation AI technology platform            aiforia.com              Aiforia launches next-generation AI technology platform    Aiforia expands its AI technology platform featuring the powerful Vision Transformer architecture and proprietary Foundation Engine.

Aiforia’s “deep learning neural networks”?                    facebook.com              1,4 milj. näyttökertaa · 23 t. reaktiota | Totally not winging it:...    Totally not winging it: these birds were trained to diagnose cancer with 99% accuracy. 🐦  In this popular Ri talk, Hannah Fry told us about a study that definitely ruffled some feathers.  Watch the...                              (A little lightheartedness for the thread )

Aiforia Technologies

Aiforia continues its successes on the customer front, but H1 revenue was well below our expectations. We see clear opportunities for the company to be one of the winners in its market in the long term. The downside to the potential is the low predictability of the growth rate also seen in H1 and the clearly elevated financial risks. At a rising valuation level, we believe that the compensation for these risks over a 12-month period is too thin. 

<p>Aiforia continues its successes on the customer front, but H1 revenue was well below our expectations. We see clear opportunities for the company to be one of the winners in its market in the long term. The downside to the potential is the low predictability of the growth rate also seen in H1 and the clearly elevated financial risks. At a rising valuation level, we believe that the compensation for these risks over a 12-month period is too thin. </p> 

	2022	23e	24e
Revenue	1.9	2.7	5.6
growth-%	91.8 %	46.5 %	103.5 %
EBIT (adj.)	-10.2	-12.1	-12.4
EBIT-% (adj.)	-546.2 %	-441.0 %	-222.3 %
EPS (adj.)	-0.41	-0.46	-0.44
Dividend	0.00	0.00	0.00
Dividend %
P/E (adj.)	neg.	neg.	neg.
EV/EBITDA	neg.	neg.	neg.

Aiforia: The journey won't be a straight line

Login required

Key Estimate Figures28.08.2023

Forum discussions